'N 8@p v /hYGGS]ʃS]ʃSSSgSgSSS}"S}"S%S`%S9S9S9~Sh9~S߃Sp߃S]S]SShSsS4sSMSMSS< SWSD WSS SmS< mSS S3DS 3DSJSpJSQSQSYZSYZSSjSS: SOSj OS SJ SSSSzS&S&S#S#SSzS5lS5lSySrySYSYS)SB")SSb$} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Pi b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1% bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Pi b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1%  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Pi  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1% bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)")B)Z-b1reasonLocalCopyvalue7bytes144patternContiguousinvocations8k bJoperation.3exchangezOperatorInputs2value\"&hTr *^ j    B?u«RM2IrB!WR/e b!"#";$"^%"&"'"(")" *"0+"S,"v-"."/"0Z"1ZE2Zh3Z4Z5Z6Z7Z8Z:9Z]:J;J<:=*> ? /@QABCDaEBFG%H*eIrJKL:PMzNOP:)Qz_RST:Uz8VnWX:YzZG[}\:]z^ _V `: a b cj6djYej|fjgjhjijjj+kjNljqmjnjojpjqj rjCsjftjujvjwjxjyj8zZ[{Z~|J}:~*  -r<R:@µ J+aʗ J:p J I  J  "  Y J    2 Jh        ;  ^       0  S  v      % H k   <_ʂJPB:Z"bQ*j3ijT*j-c*j<*sj*Ljʥ***2*U*x*****'*J*m******?*b****4 Wyڿj(jb5 jw ¹   Z2  h  ڞ    Z A w  Z  P  Z  ) ` Z  9Zo !z5"zX#{$ڞ%&'(*)M*p+ړ,ڶ-./0B1e2ڈ3ګ456778Z9}:ʠ;<= >,?O@zrABrC@DjEFGrIHʋI JJK.LdM NJOP=Q tRJSTU MVJWXY &ZJ\[\] ^J5_k`ʡa*b*c d.eQftghijk#lFminopqrs;t^uvwxy zz0{zS|jv}Z~J:*G" rT ʖ " z!Q!!:!z!*"`":"z"#9#:p#z##$:I$z$$$:"%zX%%%:%z1&T&w&:&:&:& # F i      ; ^       0 S v     $ G j  B  :[   B" d  2 r I  2 r " X 2 r  1 jh   * jA w  * j P  *   & I l      > a      3 V y     ( K n ʑ     *c  " z- o *  6 m Z   F Z|    ZU    Z. d ښ  Z = s  Z  L !:p ": # $ % & 'B (e ) * + , - .7 /Z 0} 1 2 3 4  5, 6O 7r 8 9 : ; <z! =jD >Zg ?J @: A B25 Cz D* E FA G2 H I J 3 KJi L M N  OJB Px Qʮ R  SJ TQ Uʇ V  WJ X* Y` Z [J \! ]9! ^ p! _J! `! a! b"" cJF" dJi" eJ" fJ" gJ" hJ" iJ# jJ;# kJ^# lJ# mJ# nJ# oJ# pJ $ qJ0$ rJS$ sJv$ tJ$ uJ$ vJ$ wJ% xJ%% yJH% z:k% {:% |*% }% ~ % & <& & &  ' L' 2' ' ( :V( z( ( ( :/) ze) ) ) :* z>* t* * :* z+ M+ + :+ z+ &, \, :, z, , 5- :l- - - - - . >. a. . . . . / 3/ V/ y/ / / / 0 (0 K0 n0 0 0 0 0 1 @1 c1 چ1 ʩ1 1 Z2 T2 R2 2 3 Za3 3  3 J4 R4 ʈ4  4 J4 +5 a5  5 J5 6 :6  q6 J6 6 7  J7 J7 7 7  #8 JY8 8 8  8 j9 jB9 e9 ʈ9 ʫ9 9 9 : 7: Z: }: ʠ: : :  ; ,; O; r; ʕ; ʸ; ; ; !< D< g< ʊ< ʭ< < < = 9= \= *= = "*> zl> Ү> *> 3? u? ? Z? @ N@ @ Z@ @ 'A ^A ZA A B 7B ZmB B B C ZFC |C ڲC C ZD UD ڋD !:D ":D #D $E %;E &^E 'E (E )E *E + F ,0F -SF .vF /F 0F 1F 2G 3%G 4HG 5kG 6G 7G 8G 9G :H ;=H <`H =H >zH ?jH @ZH A1I BRtI CI DI Er>J FʀJ G"J HzK I;K JqK K:K LzK ML NJL O:L PzL QL R#M S:ZM TzM UM VM W", Xbb Y Z [" \b; ]q ^ _" `b a7 bZ c"~ d" e" f" g"  h"- i"P j"s k" l" m" n" o"" p"E q"h r" s" t" u" v" w": x"] y" z" {" |" }  ~/ Q t   zB ҄ *   K 2 r  0 2g r   2@ rv   2 rO   2 r( ^  2 r 7 m 2     0 S v     % H k      = `      2 U x  Ҿ   RJ  J  V R   BT    B- c ™  B < r  B  K  B  $ [ B   4 bW bz     ) L o ’ µ    A d ‡ ª    6 Y | Ÿ    + N q  " z b r  ") zk ҭ  R P ҆  R )! _! ! R! " 8" o" R" " # H# R~# # # !$ RW$ $ $ !2$ "2 % #-% $P% %s% &% '% (% )% *"& +E& ,h& -& .& /& 0& 1' 2:' 3]' 4' 5' 6' 7' 8 ( 9/( :R( ;u( <( =( >r( ?b) @R$) Ai) BJ) C) DB4* Ev* F* GJ* H=+ Is+ J"+ Kb+ L, ML, N", Ob, P, Q%- R"\- Sb- T- U- V"5. Wbk. X. Y. Z"/ [bD/ \z/ ]/ ^"/ _b0 `S0 aw0 b0 cb0 db0 eb1 fb&1 gbI1 hbl1 ib1 jb1 kb1 lb1 mb2 nb>2 oba2 pb2 qb2 rb2 sb2 tb3 ub33 vbV3 wby3 xb3 yb3 zb3 {b4 |b(4 }RK4 ~Bn4 24 "4 4 <5 5 5 j6 H6 6 r6 7 97 2p7 r7 7 8 2I8 r8 8 8 2"9 rX9 9 9 29 r1: g: : 2: r ; @; v; 2; r; < )< 2M< 2p< 2< 2< 2< 2< 2= 2B= 2e= 2= 2= 2= 2= 2> 27> 2Z> 2}> 2> 2> 2> 2 ? 2,? 2O? 2r? 2? 2? "? ? !@ C@ @ @ A SA :A A B B]B B B C B6C lC ¢C C BD ED {D D BD E TE E BE E -F dF BF F G =G BsG G G G H #H FH iH H H H H I ;I ^I I I I I  J 0J SJ vJ J J J K %K HK jK K ҰK K bL [L ZL L  &M bhM M M R#N YN ҏN N RN 2O hO O RO  P AP xP RP P Q QQ RQ Q Q *R R`R R R S !r&S "rIS #lS $ҏS %ҲS &S 'S (T )>T *aT +҄T ,ҧT -T .T /U 03U 1VU 2yU 3ҜU 4ҿU 5U 6V 7(V 8KV 9nV :ґV ;ҴV <V =V >W ?@W @cW A2W BW C*1X DsX EڵX F2X G:Y H|Y I"Y JbY KZ LUZ M"Z NbZ OZ P.[ Q"e[ Rb[ S[ T\ U">\ Vbt\ W\ X\ Y"] ZbM] [] \] ]"] ^b&^ _\^ `^ aB^ bB^ c^ d_ eB_ fe_ g_ h_ i_ j_ k` l7` mZ` n}` o` p` q` r a s,a tOa ura va wa xa ya z!b {Db |gb }b ~b rb bb 9c Z{c  c bd Dd d zd  e Ae Rwe e e "f bOf f f 2f r'g \g g Bg g 4h kh Rh h  i "Ci byi i i 2j >j aj j j j j k 3k Vk yk k k k l (l Kl nl l l l l m @m cm m m m m n 5n Xn 7 :z2breasonOperatorInputsvalue7invocations1 -Q#h:^ 0 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1 bWuoperation.4ipu_stack_copy_u64%:%R)2))2)")B)Z-b1reasonLocalCopyvalue8bytes144patternContiguousinvocations16- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations15Y @dKo!Vz,a7lBw)M4X ? c    J n   U y   + `   6 k  Av(L3W/ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1<5jT >s(]G|1 f  1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)")B)Z-b1reasonLocalCopyvalue10bytes512patternContiguousinvocations56 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10FF  LL)D*D44>2?2IIST^^hiss}8 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations19 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 : bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5! bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1FF44 " "  (({{i i""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIZJZLLNDODQQS.T.VVXY[[]^`w`wbbdaeaggiKjKllnAoAqqs+t+vvxy{{}}tt^^ӇӉHH22"@ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1#A bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xGS;˃S;˃SSdSSlSW!SW!S$S$S:S@:SSS7S7SϐS@ϐSMS MS+S+SSSS SGS GSSd S CS  CS;TS ;TSw]S w]SSSȅSȅSSSSS'S 'SSSSpS)lSt)lSySTyS#S#S߉S$߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vB b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 I5~ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vB b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 I5~ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vB  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 I5~  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`8\o$Mql+O4X  -X|x D Tx'K b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@%I]_Bf$a_Ac|e:^ gX| ikm o   q   s   u   w  y{ &,28> D&J,P2V8\>bDhJnPtVz\  b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d%]`:VtQ. hE"  `  ?   } \ ;yX bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  PP)<*<44>'?'IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33##o o""$[%['')E*E,,.///11346689;x;x==?J@JBBD4E4GGI7J7LLNOQQSTV{V{XXZc[c]]_K`Kbbd?e?ggi'j'llnoqqssukvkxxzS{S}};;##mm  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1GS]ʃS]ʃSS|SgSgSSS}"S|}"S%SH%S9S9S3~SP3~S߃SX߃S]S]SSPSsSsSGSGSS$ SWS, WSS SmS$ mSS S3DS| 3DS/TSX/TSk]S|k]SSS…Sx…SS\SSS%S%SS SSS#lS#lSySyS S  S߉Sr߉SkS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Of b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1% bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Of b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1%  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Of  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1% bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`=aGk)Al4X^ 6Z]%ITx@ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@GkAvA(L3W_ 5Y>b /ImvLpTx"Fc9] 3 z  P t  & J   g  = a  3xJn@W{)Md6Z,qCg9~ Pt"F] b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7:operation.5exchange :*reason9binvocations1dGRU<qd+r9G Q   \ !  f  + p 5z bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10FF  KK)=*=44>'?'IIS T ^^hhrrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1FF22    ##qq] ]""$P%P''):*:,,.$/$11346688:m;m==?W@WBBDAEAGGI+J+LLNOQQSSUjVjXXZP[P]]_6`6bbdeggijlulunnpgqgssuMvMxxz3{3}}rrXXˎː>># b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xGSA˃SA˃SSdSSlSW!SW!S$S$S:S@:SSS7S7SϐS@ϐSMS MS+S+SSSS SGS GSSd SKS KS SS  SSE\S E\S!S!S[ʅS[ʅSSXSSS|SSQ(SQ(S S4 SSSlSlSySyS Sh S߉S߉SSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xD b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 J6 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xD b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 J6 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xD  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 J6  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b)goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1D8\<`-Qxb:0TRv b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b)goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1$h 0$H<`Tx:lu3W4Pt( L  2  @ d  X |   p   8,PDh\ p2"F6ZJn^ r 4$H8\Lp` t  "!!!"6""" b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dhP8 LpBX}@8( v \ B (   tZ@&  bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  LL):*:44>$?$IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%qq[ [""$I%I'')3*3,,./11346|6|88:f;f==?7@7BBD!E!GGI$J$LLN O Q}Q}SSUcVcXXZI[I]]_/`/bbdeggijlulunnpYqYssu=v=xxz!{!}}ww]]ЉЋCC))  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xMSʃSʃSSdSSlSc!Sc!S$S$S9S@9S]~S]~SCSCSېS@ېSYS YSqSqSSSS SSS SSSd SCS CSRS RSO[S O[SSS#ɅS#ɅSSSSS'S 'SSSSpS)lSt)lSySTyS5S5SS$SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations15i b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 D5 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations15i b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 D5 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations15i  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 D5  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`:u`Bih]&JpQu.Rx  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@:aq k'Ku8\Im(#Z~24k<"F,P6 Z  @ d  J n  T x  ^  hr|'5C-Q;_ImW{es,:$H2V@dN r  \!! b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1daMB#;reF'iJ-   s  V  9  bE( bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  RR)>*>44>(?(IIST^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbVV J J  >>22""  ""$p%p'')\*\,,.H/H113444668 9 ;;==?h@hBBDPEPGGIPJPLLN8O8QQS T VVXY[|[|]]_d`dbbdVeVggi>j>lln&o&qqstvvxxzj{j}}RRƂƄ::!!zz  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xMS5˃S5˃SSdSSlSc!Sc!S$S$S:S@:SSSCSCSېS@ېSYS YS%S%SSSS SSS SSSd SCS CSSS SS\S \S)S)S]S]SSSSSSSS$SSS lS lSySyS#SX#S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1k7 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 T[  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1k7 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 T[  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1k7  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 T[   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`1Ur HlFj"?Sw)Y}f<i? b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1 DVz}Fj7[6ZNr&J,e:C|*6Z~n ^Nr> b  . R   B   b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r0_p64_c0_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements1800input0Elements1800invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1< VY*~KR&zN"vJ bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  QQ)?*?44>/?/IIST^^his]s]} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccQQ ? ?  --  n n""$\%\'')J*J,,.8/8113&4&6689;;=>@y@yBBDgEgGGIWJWLLNEOEQQS3T3VVX!Y![[]^``bbdtetggibjbllnMoMqqs1t1vvxy{{}}kk݂݄OOqq  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xMSʃSʃSSdSSlSc!Sc!S$S$S9S@9SW~SW~SCSCSېS@ېSYS YSkSkSSSS SSS SSSd SCS CSSS SS]S ]SeSeSʅSʅSSS}S}S{(S {(SSSSpSlStlSySTyS S S߉S$߉S5S(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations15i b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 C5 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations15i b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 C5 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations15i  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 C5  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`6ZImX|:pDh.RPt?c2h b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@j80TLph!E|62V.R CgJ n  T x  . f e    @   4  , P  Hld2(L@dX|p 8,PDh\ t$<0THl`  !x!!"("""#@# b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1djT>()`|fNP:$  t \ D , |dL4 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  NN)6*644>?IIST^z^zhhr^s^} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbRR B B  22""j j""$R%R''):*:,,."/"113 4 6~6~88:f;f==?3@3BBDEGGIJLLNNPjQjSSUNVNXXZ2[2]]_`bbdegtgtiikXlXnnp>   b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xMS5˃S5˃SSdSSlSi!Si!S$S$S:S@:SSSISISᐄS@ᐄS_S _S%S%SSSS SYS YSSd SKS KSQS QS1[S 1[S Sh SSSCSCSgS: gSeS eS穆S:穆SSSkSkSsySsyS S S߉S߉SAS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m9 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 U[  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m9 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 U[  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m9  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 U[   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b)goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1D-Q{n$qeg/S] b-goperation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z1z5"5z9"9b=reasonOperatorInputsvalue7kernelSpecoutputElements576input0Elements576invocations1 b)goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z1z5"5z9"9b=reasonOperatorKernelvalue8kernelSpecoutputElements648input0Elements648invocations1- bJ7operation.4exchange :JbrPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })bPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } }):reasonzOperatorInputs2value8binvocations1'5Y @dKo!Vz,a7lBMqw+)MdB4X{ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r0_p64_c0_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements1800input0Elements1800invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1<5jT >?sx(1]j# bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  KK)5*544>!?!IIS T ^^hhrLsL} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaKK 5 5    ~~hhR R""$<%<'')&*&,,./11335o6o88:Y;Y==?C@CBBD-E-GGIJLLNOQxQxSSUbVbXXZL[L]]_6`6bbd e ggi j llnnpgqgssuKvKxxz/{/}}PP‰‹44  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSʃSʃSSdSSlSu!Su!S$S$S9S@9S]~S]~SUSUS퐄S@퐄SkS kSqSqSSSS SeS eSSd S+CS +CSQS QSZS ZS S SóSóSmSmSSSSSS$SSSlSlSySyS_SX_S/S/SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations16j b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 5 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations16j b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 5 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations16j  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 5  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` 1eNr3W2Vc%<hIm 6Z-Qn b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1 1X|jPt$HHl;_@dRv8\ 0i0T$H(L D< 4   ,   $ |    b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r0_p64_c0_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements1800input0Elements1800invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1< XF08`q8h@p bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  TT)F*F44>8?8IIS$T$^^h i sese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbTT F F  88**  "#%y%y'')k*k,,.]/]113O4O668A9A;;=3>3@@B%C%EEGHJJLMO~O~QQSlTlVVXZYZ[[]H^H``b4c4eeghjjlmo{o{qqsatavvxGyG{{}-~-SSŌŎ77  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lSS/˃S/˃SSLSSTS{!S{!S$S$S:S(:S S S[S[SS(SqSqSSSSSS SkS kSSL S1CS 1CS?SS ?SS{\S {\SÛSÛS-S-SSSSDSSS/lS /lSSSSS#S0#SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`, b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 _ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`, b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 _ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`,  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 _  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`*Ny'QuiKoY} -QIm{;m&J b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@7n$H5YFj W{ 1h bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1J8o } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  [[)G*G44>3?3IIST^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1gg[[ O O  CC77++  "#%%'')u*u,,.a/a113M4M668999;;=%>%@@BCEEGGIsJsLLN[O[QQSCTCVVX+Y+[[]^``bbdoeoggiWjWlln?o?qqs't'vvxy{{}}kk߂߄SSLJlj;;## b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSʃSʃSSdSSlSu!Su!S$S$S9S@9S]~S]~SUSUS퐄S@퐄SkS kSqSqSSSS SeS eSSd S+CS +CS3SS 3SSo\S o\SSSSSSSaSaS_S_SS$SSSkSkSyySyyS SX S߉S߉SeS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations16j b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 5 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations16j b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 5 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations16j  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 5  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`-Q),P}5%I|A$Hu)MAt@ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations15Y @dKo!Vz,a7lB;_w+ORv 06Zi# G  : ^  Q u  b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r0_p64_c0_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements1800input0Elements1800invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1<5jT >*sc(]U bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  PP)>*>44>,?,IIST^^hhrOsO} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$[%['')I*I,,.7/7113%4%6689;;=>@x@xBBDfEfGGITJTLLN>O>QQS(T(VVXY[[]]_q`qbbdWeWggi=j=lln!o!qqstvtvtxxzV{V}}88ssUUƎƐ77  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lSS/˃S/˃SSLSSTS{!S{!S$S$S:S(:S S S[S[SS(SqSqSSSSSS SkS kSSL SKS KSeQS eQSZS ZS}S}S]S]SIS@ISSSSSlSlS St S߉S߉SSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b. b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ` bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b. b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ` bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b.  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 `  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b)goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1D"FiZ~;_9] 1#vbHl b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b)goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1$By/S @d bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1&]K } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  UU)=*=44>%?%IIS T ^{^{hhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%% } }""$i%i'')Q*Q,,.9/9113!4!668 9 ;};}==?e@eBBDMEMGGI5J5LLNOQQSSUoVoXXZS[S]]_7`7bbdeggiikqlqnnpUqUssu9v9xxz{}}ssWWɉɋ;; b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lSSʃSʃSSLSSTS!S!S%S%S9S(9Sc~Sc~SgSgSS(S}S}SwSwSSSS SwS wSSL S=CS =CSQS QS1[S 1[SySyS]S]SSS)SD)SSS;lS ;lSSS_S0_SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations17k b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 97  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations17k b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 97  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations17k  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 97   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`(<`xd>nBf  Ddg"Z~Ei b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@(c!E>b [=x6Z bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1?z5p } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  RR)B*B44>2?2IIST^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbRR B B  22"" z z""$j%j'')Z*Z,,.J/J113:4:668*9*;;=>@@B C EEGGIrJrLLN^O^QQSJTJVVX6Y6[[]"^"``bceeggipjplln\o\qqsHtHvvx4y4{{} ~  nnZZБГ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lSS)˃S)˃SSLSSTS!S!S%S%S:S(:S~S~SsSsS S( SSSSSwSwSS SS SSL SICS ICSRS RS'\S '\SoSoSSSSS SD SSSlS lS S S߉S0߉SSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U! b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 j[. bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U! b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 j[. bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U!  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 j[.  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Hl#Gn D6Z7dNr<`Cf] b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@*_5j@u'K2V=a bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1;p%Z } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  OO)9*944>!?!IIST^z^zhhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((tt^ ^""$H%H'')2*2,,./11346{6{88:e;e==?M@MBBD3E3GGIJLLNNPrQrSSUXVXXXZ>[>]]_$`$bbd e g}g}iikclcnnpIqIssu/v/xxz{}}nnTTƉƋ88 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lSSʃSʃSSLSSTS!S!S$S$S9S(9Sc~Sc~SaSaSS(SwSwSwSwSSSS SqS qSSL S7CS 7CSRS RS[S [SSSKSKSaSaSSDSSS#lS #lSSS߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations17k b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 86 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations17k b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 86 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations17k  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 86  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$Hbn$0THl"$Hel#5YIm#Gm b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Sw5p.R K o - h & J  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1+f!\ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  NN):*:44>&?&IIST^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&tt` `""$L%L'')8*8,,.$/$11346688:r;r==?^@^BBDJEJGGI6J6LLNOQQSTVzVzXXZb[b]]_J`Jbbd2e2ggijllnoqvqvssu^v^xxzF{F}}..rrZZ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lSS)˃S)˃SSLSSTS!S!S%S%S:S(:SSSsSsS S( SSSSS}S}SS SS SSL SKS KSyRS yRS[S [SSS'S'S1S@1SSSSSlSlS{ St{ SK߉SK߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W# b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 k[0 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W# b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 k[0 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W#  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 k[0  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b)goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1D;Bf:8u,P7[BfX| b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b)goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1$HlSw)^ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1Dy.c } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  MM)3*344>?IISS]j^jhhrLsL} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((ppV V""$<%<'')"*",,./1{1{335a6a88:G;G==?+@+BBD E G~G~IIK`L`NNPBQBSSU$V$XXZ[]w]w__aYbYddf;g;iiklnnpprpspuuwRxRzz|4}4hh؋؍HH b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLSSTS!S!S)%S)%S:S(:Si~Si~SSS#S(#SSS}S}SSSS SS SSL SaCS aCSQS QSZS ZS7S7SíSíSSS#SD#SSS5lS 5lSYSYS)S0)SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations18l b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  M@ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations18l b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  M@ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations18l  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  M@  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$/Ssf5Uy5Y7[;d=Fj b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$f2V"d0T b.R bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1BF } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  PP)>*>44>,?,IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$[%['')I*I,,.7/7113%4%6689;;=>@x@xBBDfEfGGITJTLLN>O>QQS(T(VVXY[[]]_q`qbbd[e[ggiEjElln/o/qqstvvxy{x{x}}bbׂׄLL66  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS#˃S#˃SSLSSTS!S!S %S %S:S(:S~S~SmSmSS(SSS S SqSqSS S}S }SSL SCCS CCSRS RSU[S U[SSS S SSSSDSSS/lS /lS)S)S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uJ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 j" bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uJ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 j" bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uJ  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 j"  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`5Y@\  0Bf[ /MqZ~=Y} b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@(]3h>s%I~0T;_ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations19n#X } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  TT)>*>44>)?)IIST^^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggWW G G  77''s s""$]%]'')G*G,,.1/111346689;z;z==?d@dBBDNENGGI9J9LLN!O!QQS T V}V}XXZe[e]]_M`Mbbd5e5ggijllnoqyqyssuavaxxzI{I}}11yycc b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLS ST S!S!S#%S#%S:S(:Si~Si~SSSS(SSS}S}SSSS SS SSL S[CS [CSRS RS=[S =[SSS#SZ#SS:SSjSS S#lS* #lS S  S߉S߉SkS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations18l b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  L> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations18l b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  L> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations18l  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  L>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`?fX|4XAZ~'Cj\8\!E^ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z1z5*5z9*9b=reasonOperatorKernelvalue8kernelSpecoutputElements1152input0Elements1152invocations1bPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@`,P^* N  \ ( L  Z ~  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1J NR } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  OO)9*944>#?#IIS T ^|^|hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''uua a""$L%L'')6*6,,. / 113 4 6688:i;i==?S@SBBD=E=GGI'J'LLN O QQSSUfVfXXZL[L]]_2`2bbdeggiikqlqnnpWqWssu=v=xxz#{#}} ||bbՉՋHH.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS#˃S#˃SSLSSTS!S!S %S %S:S(:S~S~SmSmSS(SSSSSwSwSS S}S }SSL SKS KSkQS kQSZS ZSSS'S'S1S@1SSSSS#lS#lS St S߉S߉SYS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vL b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 k$ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vL b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 k$ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vL  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 k$  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b)goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1D 0p0T{>b&e2+OuCg b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b)goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1$FjSw)^ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 Bw,a } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  PP):*:44>$?$IIS T ^{^{hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))uu_ _""$I%I'')3*3,,./11346|6|88:f;f==?P@PBBD:E:GGI$J$LLN O Q}Q}SSUcVcXXZI[I]]_/`/bbdeggiikilinnpMqMssu1v1xxz{}}kk݄݆QQĉċ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTSu!Su!S$S$S :S( :So~So~SUSUS퐄S(퐄SkSkSSSSSS SeS eSSL S+CS +CSqQS qQSZS ZSSSqSqSSS#SD#SSS5lS 5lS;S;S S0 SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19m b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 .5  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19m b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 .5  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19m  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 .5   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$6Z'Kq' 1lSw>b9]f4?Lp b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$qSw5Y;jLp.R bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1Mg4 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  VV)B*B44>,?,IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffZZ N N  BB66&&  ""$t%t'')`*`,,.L/L113848668$9$;;=>@@BBDlElGGITJTLLN3?3IIS!T!^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggUU C C  11  r r""$`%`'')N*N,,.@}@}BBDkEkGGI[J[LLNIOIQQS7T7VVX%Y%[[]^``bcexexggifjfllnOoOqqs3t3vvxy{{}}mm߂߄QQÇÉ55 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTSu!Su!S$S$S :S( :So~So~SUSUS퐄S(퐄SkSkSSSSSS SeS eSSL S+CS +CSQS QS7[S 7[SSSSS S S SD SSSlS lS S S߉S0߉SAS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19m b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 -5  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19m b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 -5  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19m  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 -5   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`6HlBf<<Hl BTxNr$H$HTx b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@4cE i ' K - z  \ > b  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1Nh5 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  RR):*:44> ? IIST^v^vhhrZsZ} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&&n n""$V%V'')>*>,,.&/&11346688:j;j==?P@PBBD4E4GGIJLLNNPnQnSSURVRXXZ6[6]]_`bbddfpgpiikTlTnnp8q8ssuvxxz{}r}rVVȄȆ:: b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S˃S˃SSLSSTSu!Su!S$S$S:S(:S~S~SUSUS퐄S(퐄SkSkSSSkSkSS SeS eSSL SKS KSeQS eQSZS ZS}S}SϰSϰSmS@mSSSSS)lS)lS St S߉S߉SqS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kA  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 R` bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kA  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 R` bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kA   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 R`  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b)goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1D%%Ijn"8\l*mDh;_ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b)goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1$$` D@d $  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1,h$` } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  OO)9*944>%?%IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeOO 9 9  ##  llV V""$@%@'')***,,./11335s6s88:];]==?G@GBBD1E1GGIJLLNOQ|Q|SSUfVfXXZP[P]]_:`:bbd$e$ggijllnnpgqgssuKvKxxz/{/}}iiۉۋMM11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTS{!S{!S$S$S:S(:Su~Su~S[S[SS(SqSqSSSSSS SkS kSSL S1CS 1CSQS QSZS ZS7S7SSS7S7S5SD5S S SGlS GlSSS_S0_SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1:n b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 6  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1:n b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 6  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1:n  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 6   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`5Y&&J~7Bv:^n:^2f b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$|tl d\TxLpDh bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1X0` } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  XX)J*J44>7@@B)C)EEGHJJL M OOQQSpTpVVX^Y^[[]L^L``b8c8eeg"h"jjl m ooqqsetevvxKyK{{}1~1ooSSőœ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S˃S˃SuSLuSSTS{!S{!S$S$S:S(:S~S~S[S[SS(SqSqSSSeSeSS SkS kSSL S1CS 1CSkQS kQSZS ZSSSSSSS/SD/SSSAlS AlSeSeS5S05SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_4 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 V  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_4 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 V  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_4  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 V   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`5Y28\u(L^Rvv*NNrrDh b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$kAe;X|.R(oEi bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1GU } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  __)K*K44>7?7IIST^^his{s{} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kk__ S S  GG;;//##  "#%%'(*y*y,,.e/e113Q4Q668=9=;;=)>)@@BCEEGHJwJwLLN_O_QQSGTGVVX/Y/[[]^``bbdsesggi[j[llnCoCqqs+t+vvxy{{}}ooWWˇˉ??'' b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTS!S!S$S$S :S( :So~So~SaSaSS(SwSwSSSSSS SqS qSSL S7CS 7CSQS QS7[S 7[SSSնSնSoSoS#SD#SSS5lS 5lS#S#S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1:n b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 6 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1:n b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 6 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1:n  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 6  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`5Y -/S"9g9]9]CCg /)MMq b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@<`4 X , P $ H  @  8 0( bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations18h@ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  TT)B*B44>0?0IIST^^hhrlsl} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffTT B B  00  q q""$_%_'')M*M,,.;/;113)4)6689;;=>@|@|BBDjEjGGIXJXLLNBOBQQS,T,VVXY[[]^`u`ubbd[e[ggiAjAlln%o%qqstvxvxxxzZ{Z}}<<rrTT b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S˃S˃S{SL{SSTS{!S{!S$S$S:S(:S~S~S[S[SS(SqSqSSSeSeSS SkS kSSL SKS KSkQS kQSZS ZSSSݱSݱSS@SSSSS)lS)lS St S߉S߉SeS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`6 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 U  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`6 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 U  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`6  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 U   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b)goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1D$$Hk<`2 .qm?c b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b)goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1$?\2 V  ,  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1c*q8 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  YY)A*A44>)?)IIS T ^^hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiYY I I  99))   ""$m%m'')U*U,,.=/=113%4%668 9 ;;==?i@iBBDQEQGGI9J9LLNOQQSTVsVsXXZW[W]]_;`;bbdeggijlulunnpYqYssu=v=xxz!{!}}ww[[͉͋??## b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSSLSSTS!S!S$S$S:S(:S{~S{~SaSaSS(SwSwSSSSSS SqS qSSL S7CS 7CSRS RSC[S C[SSSSSgSgS;SD;SSSMlS MlSSSwS0wSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 #5 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 #5 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 #5  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`5Y&&J"4g4X9]BBf /,PPt b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$2@*N8\FjTxb  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1cF)  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  VV)F*F44>6?6IIS"T"^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&& ~ ~""$n%n'')^*^,,.N/N113>4>668.9.;;=>@@BCEEGGIvJvLLNbObQQSNTNVVX:Y:[[]&^&``bceeggitjtlln`o`qqsLtLvvx8y8{{}$~$rr^^ԑԓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS ˃S ˃SoSLoSSTS!S!S$S$S:S(:S~S~SaSaSS(SwSwSSS_S_SS SqS qSSL S7CS 7CS}QS }QSZS ZSSSSS%S%SSDSSS/lS /lS S S߉S0߉SkS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T) b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Yj bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T) b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Yj bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T)  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Yj  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`++Ol~ @d,PPtz6{QuvMqMq b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$vbNr:^&J6"t bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1R$vH } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  SS)=*=44>%?%IIS T ^~^~hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,xxb b""$L%L'')6*6,,. / 113 4 6688:i;i==?Q@QBBD7E7GGIJLLNOQvQvSSU\V\XXZB[B]]_(`(bbdeggiikglgnnpMqMssu3v3xxz{}}rrXXʉʋ<<  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSSLSSTS!S!S%S%S:S(:Su~Su~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CSQS QS1[S 1[SySyS1S1SSS)SD)SSS;lS ;lS/S/S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 "5 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 "5 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 "5  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`5Y''K$ 11Ui;_;;_ 1!EEi b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@p ~  )  7 ! E / S =aKoY} bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1oR5{ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  RR)>*>44>*?*IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **xxd d""$P%P'')<*<,,.(/(11346689;v;v==?b@bBBDNENGGI:J:LLN"O"QQS T V~V~XXZf[f]]_N`Nbbd6e6ggijllnoqzqzssubvbxxzJ{J}}22vv^^ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS ˃S ˃SoSLoSSTS!S!S$S$S:S(:S~S~SaSaSS(SwSwSSS_S_SS SqS qSSL SKS KSQS QSZS ZSSS岅S岅S?S@?S S SSSlSlS St S]߉S]߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U+ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 X k bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U+ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 X k bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U+  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 X k  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b)goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1D$$HoCg D9@~ <Pt b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b)goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1$`Lp8 \ $ H  4  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1l>b } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  QQ)7*744>?IISS]n^nhhrPsP} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,ttZ Z""$@%@'')&*&,,. / 11335e6e88:K;K==?/@/BBDEGGIIKdLdNNPFQFSSU(V(XXZ [ ]{]{__a]b]ddf?g?iik!l!nnpqststuuwVxVzz|8}8ll܋܍LL b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS˃S˃S{S{SS$S:S:S~S~SUS$USSSkSkSS$SqQSqQSZS4ZSST S_ S _ SGS GSS SYlS YlSS SSSSSS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1EyM3 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1EyM3 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1EyM3 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 GkA^4X  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ww  ee)S*S44>???IIS)T)^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1wwee S S  AA//   ""$p%p'')^*^,,.L/L113:4:668(9(;;=>@@BCE{E{GGIgJgLLNQOQQQS;T;VVX%Y%[[]^``bbdnenggiXjXllnBoBqqs,t,vvxy{{}~uu__ԇԉII33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS˃S˃SiS iSSS}:S}:S~S~SmSmSSpSYSYSSSQS`QSZSZSS$ S_ S _ S;S ;SS SMlS MlSGS GSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 i;_ 1vHl  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  aa)K*K44>4?4IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ttdd T T  DD44!!   ""$j%j'')T*T,,.>/>113(4(6689;;==?q@qBBD[E[GGIDJDLLN,O,QQSTVVXXZp[p]]_X`Xbbd@e@ggi(j(llnoqqssulvlxxzT{T}}<<$$ ll b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS˃S˃SSSS$S:S:S~S~SUS$USSSkSkSS$S;QS;QSwZS4wZSST S_ S _ S5S 5S S  SGlS GlSS S߉S߉SS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1FzK6 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1FzK6 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1FzK6 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 @d,n6Z'  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10vv  bb)L*L44>4?4IIST^^hissss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvbb N N  ::&&t t""$_%_'')I*I,,.3/311346689;|;|==?f@fBBDPEPGGI8J8LLNOQQSTVwVwXXZ][]]]_C`Cbbd)e)ggijllnnphqhssuNvNxxz4{4}}ssYY̎̐?? b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS˃S˃SiS iSSS}:S}:S~S~SmSmSSpSYSYSSSQS`QSZSZSS$ S_ S _ S/S /SS SAlS AlS S  S߉S߉SwS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 qCg9~ Pt  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  __)I*I44>1?1IIST^^hhrlsl} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1tt`` L L  88$$n n""$X%X'')B*B,,.,/,11346689;u;u==?_@_BBDIEIGGI1J1LLNOQQSSUpVpXXZV[V]]_<`v>xxz"{"}}xx\\ωϋBB(( b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSS SSS):S):S~S~SySySSpSSSSSqQS`qQSZSZSS$ SY S Y S;S ;SS SMlS MlSSS SS#S#SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1#} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1#} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1# k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 i7[)l;_  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ww  gg)S*S44>;?;IIS#T#^^h i s}s}} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1wwkk _ _  SSGG77##  "#%%'')q*q,,.]/]113I4I668595;;=>@@BCE{E{GGIcJcLLNKOKQQS3T3VVXY[[]^`w`wbbd_e_ggiGjGlln/o/qqstvvxxzs{s}}[[ςτCC(( b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS ˃S ˃SoS oSSS:S:S~S~S=S=SSpS_S_SSS}QS`}QSZSZSS$ SY S Y S;S ;SS SMlS MlSeS eS5S5SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q& b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1j} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q& b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1j} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q& b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1j k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 i;_ 1vHl  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  ^^)L*L44>:?:IIS(T(^^h i s~s~} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp^^ L L  ::(( { {""$i%i'')W*W,,.E/E113343668!9!;;=>@@BBDtEtGGIbJbLLNPOPQQS>T>VVX,Y,[[]^``bceeggimjmllnVoVqqs:t:vvxy{{}~ttXXʇʉ<<  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSS SSS#:S#:S~S~SsSsSSpSSSSSQS`QSZSZSS$ SY S Y S#S #SS S5lS 5lS S  S߉S߉SSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1"} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1"} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1" k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 rDh: Qu  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10uu  aa)I*I44>-?-IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1uuee U U  EE55!!  } }""$e%e'')M*M,,.5/511346689;y;y==?]@]BBDAEAGGI%J%LLN O Q{Q{SSU_V_XXZC[C]]_'`'bbd e g}g}iikalannpEqEssu)v)xxz { }}ccՄՆEE''  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS ˃S ˃SoS oSSS:S:S~S~S=S=SSpS_S_SSSQS`QSZSZSS$ SY S Y S)S )SS S;lS ;lSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P& b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1k} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P& b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1k} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P& b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1k k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 rFj>X|  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  ZZ)D*D44>.?.IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ppZZ D D  ..wwa a""$K%K'')5*5,,./113 4 6~6~88:h;h==?R@RBBDG?GIIS3T3^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ssee W W  II;;--  "#%%'(*|*|,,.n/n113`4`668R9R;;=D>D@@B6C6EEG(H(JJLMOOQRT{T{VVXiYi[[]W^W``bCcCeeg-h-jjlmooqqsptpvvxVyV{{}<~<""xx\\ΑΓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS˃S˃SuS uSSS:S:S~S~SmSmSSpSeSeSSS}QS`}QSZSZSS$ SS S S S5S 5S S  SGlS GlSqS qSASASS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z/ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z/ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z/ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 i;_ 1vHl  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  dd)P*P44>:?:IIS"T"^^h i s~s~} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ppdd X X  LL@@44((  "#%%'(*~*~,,.j/j113V4V668B9B;;=.>.@@BCEEGHJzJzLLNbObQQSJTJVVX2Y2[[]^``bcevevggi^j^llnFoFqqs.t.vvxy{{}}rrZZ·ΉBB** b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS S  SSS:S:S~S~S[S[SSpSSSSSQS`QSZSZSS$ SS S S S/S /SS SAlS AlS)S )S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 sQu/S3~  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  __)M*M44>9?9IIS!T!^^hissss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qq__ M M  ;;)) | |""$j%j'')X*X,,.F/F113444668"9";;=>@@BBDuEuGGIaJaLLNKOKQQS5T5VVXY[[] ^ `~`~bbddedggiJjJlln.o.qqstvvxxzc{c}}EE''yy[[ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS˃S˃SuS uSSS:S:S~S~SgSgSSpSeSeSSSQS`QSZSZSS$ SS S S S#S #SS S5lS 5lS S  S߉S߉SqS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y/ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y/ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y/ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 qCg9~ Pt  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  ``)H*H44>.?.IIST^^hhrhsh} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp`` P P  @@00    "#%t%t'')\*\,,.D/D113,4,6689;;==?p@pBBDXEXGGI>J>LLN"O"QQSTVxVxXXZ\[\]]_@`@bbd$e$ggijlzlznnp^q^ssuBvBxxz&{&}} ||``҉ҋDD(( b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃS S  S S S:S:S{~S{~SSSSpSSSSSeQS`eQSZSZSS$ SS S S SAS ASS SSlS SlSS SwSwSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1:} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1:} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1: k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 h6Z(k9]  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10oo  __)O*O44>=?=IIS)T)^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1oo__ O O  ??//  ""$w%w'')g*g,,.W/W113G4G668797;;='>'@@BCEEGHJ}J}LLNiOiQQSUTUVVXAYA[[]-^-``bceeghj{j{llngogqqsStSvvx?y?{{}+~+yyeeۑۓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S˃S˃S{S {SSS:S:S~S~S%S%SSpSkSkSSSQS`QS[S[SS$ SS S S SS SS S)lS )lS S  S߉S߉S_S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c8 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1[} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c8 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1[} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c8 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1[ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 UyCg1UC  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  TT)>*>44>$?$IIS T ^}^}hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiUU A A  --yyc c""$M%M'')7*7,,.!/!113 4 6688:j;j==?P@PBBD6E6GGIJLLNOQuQuSSU[V[XXZA[A]]_'`'bbd e ggiikflfnnpLqLssu2v2xxz{}}qqUUljNj99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃSS S S S:S:S{~S{~SSSSpSSSSSQS`QSZSZSS$ SS S S S)S )SS S;lS ;lS/S /S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1;} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1;} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1; k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 rMq'K%n  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  YY)E*E44>/?/IIST^^hhrsss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mmYY E E  11  k k""$W%W'')C*C,,.///11346689;};}==?i@iBBDUEUGGI?J?LLN'O'QQSTVVXXZk[k]]_S`Sbbd;e;ggi#j#lln o qqssugvgxxzO{O}}77{{cc b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S˃S˃S{S{SS:S:S*:S~S:~S%Sb%SSRSkSb kSS S;QSz ;QSwZS wZSS2SS S S SSSSSlSlS Sz SW߉SW߉SS: zinput1Elements} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b8 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1[} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b8 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1[} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b8 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1[ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 @d,l4X .  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  TT):*:44>?IISS]o^ohhrOsO} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kkWW C C  //ww] ]""$C%C''))*),,./11335h6h88:N;N==?0@0BBDEGGIIKeLeNNPGQGSSU)V)XXZ [ ]|]|__a^b^ddf@g@iik"l"nnpqsusuuuwWxWzz|9}9kkۋۍKK b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃSS SSS:S:Su~Su~S[S[SSpSSSSSQS`QSZSZSS$ SM S M S/S /SS SAlS AlSeS eS5S5SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 mCg=Z~  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  YY)G*G44>3?3IIST^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kkYY G G  55##v v""$d%d'')R*R,,.@/@113.4.6689;;= > @@BBDoEoGGI[J[LLNEOEQQS/T/VVXY[[]^`x`xbbdbebggiLjLlln6o6qqs t vvx y {{}}iiނބSSȇȉ=='' b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S˃S˃SS SSS:S:S~S~SaSaSSpSkSkSSSQS`QSZSZSS$ SM S M SS SS S/lS /lS)S )S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1lA  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1lA  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1lA  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Lp,P 0}]  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  UU)?*?44>(?(IIST^^hhrlsl} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhXX H H  88((t t""$^%^'')H*H,,.2/211346689;{;{==?e@eBBDOEOGGI8J8LLN O QQSTV|V|XXZd[d]]_L`Lbbd4e4ggijllnoqxqxssu`v`xxzH{H}}00vv`` b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃSS SSS:S:Su~Su~SUSUSSpSSSSSQS`QSZSZSS$ SM S M SS SS S)lS )lS S  S߉S߉SqS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 sQu/S 1|  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  VV)@*@44>(?(IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjVV B B  ..||h h""$S%S'')=*=,,.'/'11346688:p;p==?Z@ZBBDDEDGGI,J,LLNOQQSSUkVkXXZQ[Q]]_7`7bbdeggijlvlvnnp\q\ssuBvBxxz({(}}ggډڋMM33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S˃S˃SS SSS:S:S~S~SaSaSSpSkSkSSSQS`QSSZSSZSS$ SM S M SS SS S#lS #lS S  S߉S߉S_S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kA  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kA  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kA  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 5YRv D  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  SS)=*=44>%?%IIS T ^|^|hhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,xxb b""$L%L'')6*6,,. / 113 4 6688:i;i==?S@SBBD=E=GGI%J%LLN O Q~Q~SSUdVdXXZJ[J]]_0`0bbdeggiikjljnnpNqNssu2v2xxz{}}llބކPPÉË66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYSʃSʃSS SSS :S :So~So~SySySSpSSSSSQS`QS1[S1[SS$ SG S G SS SS S/lS /lS;S ;S S SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations17} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations17} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations17 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Z~VzRvPt  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  [[)G*G44>/?/IIST^^hhrqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kk__ S S  GG;;++  "#%y%y'')e*e,,.Q/Q113=4=668)9);;=>@@BBDoEoGGIWJWLLN?O?QQS'T'VVXY[[]]_k`kbbdSeSggi;j;lln#o#qqs t vvxxzg{g}}OOÂÄ77uu b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYS#˃S#˃SS SSS:S:S~S~SSS Sp SqSqSSSYQS`YQSZSZSS$ SG S G SS SS S/lS /lSMS MSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uJ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1[} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uJ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1[} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uJ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1[ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 GkA^4X  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  RR)@*@44>.?.IIST^^hisrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddRR @ @  ..  o o""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGIVJVLLNDODQQS2T2VVX Y [[]^``bbdsesggiajallnJoJqqs.t.vvxy{{}}hhڂڄLL00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYSʃSʃSS SSS:S:Si~Si~SySyS}Sp}SSSSSMQS`MQSZSZSS$ SG S G SS SS SlS lS S  S߉S߉S;S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1<p b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations18} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1<p b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations18} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1<p b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations18 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ei; Rv$H  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  UU)=*=44>!?!IIST^w^whhrYsY} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiYY I I  99))q q""$Y%Y'')A*A,,.)/)11346688:m;m==?Q@QBBD5E5GGIJLLNNPoQoSSUSVSXXZ7[7]]_`bbddfqgqiikUlUnnp9q9ssuvxxz{}s}sWWɄɆ99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYS#˃S#˃SS SSS:S:S~S~SSS Sp SqSqSSS;QS`;QSwZSwZSS$ SG S G SS SS S#lS #lS S  S߉S߉SkS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1tJ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1[} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1tJ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1[} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1tJ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1[ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 5Y+pCg9  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  NN)8*844>"?"IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddNN 8 8  ""  kkU U""$?%?''))*),,./11335r6r88:\;\==?F@FBBD0E0GGIJLLNOQyQySSUcVcXXZM[M]]_7`7bbd!e!ggi j llnnpdqdssuHvHxxz,{,}}ff؉؋JJ.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LSSʃSʃSS SSS:S:Si~Si~SOSOS}Sp}SSSSSQS`QS[S[SS$ SG S G S)S )SS S;lS ;lSS SSSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 UyGk9]+O  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  YY)K*K44>;?;IIS'T'^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggYY K K  ==//!!  "#%~%~'')p*p,,.b/b113T4T668F9F;;=8>8@@B*C*EEGHJJL M OOQQSoToVVX]Y][[]K^K``b7c7eeg!h!jjl m ooqqsftfvvxLyL{{}2~2nnRRđē b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LSS#˃S#˃SS SSS:S:S~S~SUSUSSpSwSwSSSQS`QSAZSAZSS SG S| G S#S #SK~S K~S)NS )NS_S _SUމSUމSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~S b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~S b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~S b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1Dh4x"x  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  XX)D*D44>.?.IIST^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddXX L L  @@44((  "#%%'')r*r,,.^/^113J4J668696;;=">"@@BCEEGGInJnLLNVOVQQS>T>VVX&Y&[[]^``bbdjejggiRjRllnDoDqqs,t,vvxy{{}}ppXẊ̉@@(( b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LSSʃSʃSS SSS9S9Sc~Sc~SISISwSpwSSSSS)QS`)QSeZSeZSS$ SG S G SS SS S)lS )lSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19m b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19m b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19m b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 :^a3W)  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>-?-IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIUJULLN?O?QQS)T)VVXY[[]]_r`rbbdXeXggi>j>lln&o&qqstvyvyxxz[{[}}==qqSS b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LSS#˃S#˃SS SSS:S:S~S~SUSUSSpSwSwSSSKSS`KSS\S\SS SG S| G S S  S3~S 3~SNS NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}S b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}S b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}S b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations15Y3W1U)M  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  TT)<*<44>"?"IIST^^hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$ | |""$h%h'')P*P,,.8/8113 4 6689;|;|==?d@dBBDLELGGI2J2LLNOQQSSUlVlXXZP[P]]_4`4bbdeggiiknlnnnpZqZssu>v>xxz"{"}}xx\\Ή΋@@$$ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LSSʃSʃSS SSS9S9Sc~Sc~SmSmSwSpwSSSSS#QS`#QS_ZS_ZSS SA S| A S5S 5S]~S ]~S;NS ;NSS SމSމS9S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations18l b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations14} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations18l b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations14} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations18l b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations14 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1Pt4X D(  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  SS)E*E44>3?3IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33## { {""$m%m'')]*],,.M/M113=4=668-9-;;=>@@B C EEGGIsJsLLN_O_QQSKTKVVX7Y7[[]#^#``bceeggiqjqllnioiqqsUtUvvxAyA{{}-~-{{ggݑݓ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LSS)˃S)˃SS SSS:S:SSSSSSpS}S}SSSRS`RS[S[SS SA S| A SS S9~S 9~SNS NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\( b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1[} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\( b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1[} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\( b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1[ k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1Jn9]  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10]]  RR)<*<44>"?"IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1]]II 5 5  !!  wwa a""$K%K'')5*5,,./113 4 6~6~88:h;h==?N@NBBD4E4GGIJLLNOQsQsSSUYVYXXZ?[?]]_%`%bbdeggiikplpnnpVqVssu'?'IIST^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%ss_ _""$O%O'');*;,,.'/'11346688:u;u==?a@aBBDMEMGGI7J7LLNOQQSTV{V{XXZc[c]]_K`Kbbd3e3ggijllnoqqssukvkxxzS{S}};;## gg b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LSS)˃S)˃SS SSS:S:SSSSSSpS}S}SSSQS`QSZSZSS SA S| A SS S'~S '~SNS NS S  Sw݉Sw݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\( b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1[} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\( b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1[} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\( b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1[ k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1~ zvr  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  PP)6*644>?IIST^w^whhrWsW} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__KK 7 7  ##ssY Y""$?%?'')%*%,,. / 1~1~335d6d88:J;J==?,@,BBDEGGIIKaLaNNPCQCSSU%V%XXZ[]x]x__afbfddfHgHiik*l*nnp q s}s}uuw_x_zz|A}A##ssSSÐÒ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations1MScʃScʃSS|SgSgSSS}"S|}"S%SH%S9S9S9~SP9~S߃SX߃S]S]SSPSsSsSMSMSS$ SWS, WSS SmS$ mSS S3DS| 3DSRSXRS[S|[SESESaSxaSS\SSS$S$S#S8#SSS5lS5lSySyS_Sl_S/S/SS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Pk b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1' bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Pk b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1'  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Pk  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1' bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`1U7[v4?ce#GkDh3WOsx6g b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@7[Hl"Y}3j D{1U (Bf]3Swh ->ds > % I u ~  + O 0 T     = a    P t   * a  ;r(L9]Jn$[5l"F}3W Dh b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d7n%\$JY8Cox&]K   9  p  ' ^   L bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10GG  UU)K*K44>7?7IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1GG;; / /  ==11%%  " # %%'(*y*y,,.e/e113Q4Q668=9=;;=)>)@@BCEEGHJwJwLLN_O_QQSGTGVVX/Y/[[]^``bbdsesggi[j[llnOoOqqs7t7vvxy{{}~{{ccׇ׉KK33# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xMS;˃S;˃SSdSSlSW!SW!S$S$S:S@:SSS7S7SϐS@ϐSMS MS%S%SSSS SGS GSSd S CS  CSSS SS\S \SSS]ŅS]ŅSSSSS&S &S)S)SSpS;lSt;lSySTySSSYS$YSS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1r> b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 G1~ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1r> b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 G1~ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1r>  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 G1~  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`?ckAeHl"=a;Z~j D9]| b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Rv>b*N:&xdPt<`(L 8   $ v   b  N r  : ^  &J98b(@d|Bl \kLpIm<`'K,P@ 0 tdTxDh4X$H b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dR$vHl>b4! l \ 7 0   O  X  ,  T(| bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  VV)J*J44>@@BCEEGHJJL M OOQQSpTpVVX^Y^[[]L^L``b8c8eeg"h"jjl m ooqqsqtqvvxWyW{{}=~=## {{__ёѓ  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1MS]ʃS]ʃSS|SgSgSSS}"S|}"S%SH%S9S9S9~SP9~S߃SX߃S]S]SSPSsSsSMSMSS$ SWS, WSS SmS$ mSS S3DS| 3DSRSXRS[S|[SSSąSxąSS\SSS&S&S S8 SSSlSlSySyS Sl S߉S߉S_S } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Oh b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1' bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Oh b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1'  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Oh  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1' bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`BRvp(>bFj*N-Qo:)W{-Q b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Rv>b*N:&xdPt<`(L 8   $ v   b ; _   N r F j   : ^ Q u  &J'\42g=lrVz>b&J2jSw;_#G /gOs7[C b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dR$vHl>b-4b  T L $   t  D   d 4T bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10FF  QQ)A*A44>)?)IIS T ^^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1FF66 & &  11!! y y""$m%m'')U*U,,.=/=113%4%668 9 ;;==?i@iBBDQEQGGI9J9LLNOQQSTVsVsXXZW[W]]_;`;bbdeggijlulunnpeqessuIvIxxz-{-}}ggىًKK//# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xMS;˃S;˃SSdSSlSW!SW!S$S$S:S@:SSS7S7SϐS@ϐSMS MS+S+SSSS SGS GSSd S CS  CSRS RS[S [S3S3SDžSDžSYSYSuSuSs'S s'SSSSpS)lSt)lSySTySSS߉S$߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t@ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 H2 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t@ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 H2 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t@  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 H2  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ci(L9]Nr4Sw*Nt+(L b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@]_ace g ikm o   q   s   u   w  y{ {ywus)qtoRv0Tm2k}i[ 9]g ;eca_]  b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d]:tQ. hE"}  X  3    i e D 0{zFU0 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  PP)B*B44>0?0IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$_%_'')M*M,,.;/;113)4)6689;;==?a@aBBDOEOGGIXJXLLNBOBQQS,T,VVXY[[]^`u`ubbd[e[ggiAjAlln1o1qqstvvxxzf{f}}HH** ~~``  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xMSʃSʃSSdSSlSc!Sc!S$S$S9S@9S]~S]~SCSCSېS@ېSYS YSqSqSSSS SSS SSSd SCS CSRS RS[S [SSS˅S˅SSSSS(S (SSSSpSlStlSySTyS S S߉S$߉SSS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations17k b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 @7 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations17k b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 @7 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations17k  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 @7  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`.Rp)(X|3.R=8Vz#G=a b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@l @<`\| 0,PLp l    @  w < ` - Q   \ > b   |Os 0)`,P:qLpn$$HHll""FFjj  DDhh B  B!f!!!f"""###$@$ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dlXD0t`:Lq8 (  & _      p ^ L:( bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10^^  II)3*344>?IIS T ^~^~hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1^^JJ 6 6  ""nnX X""$B%B''),*,,,./11346u6u88:_;_==?-@-BBDEGGIJLLNNPlQlSSURVRXXZ8[8]]_`bbdeggiikglgnnpMqMssu3v3xxz{}}rrXXʉʋ<<   b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xMS5˃S5˃SSdSSlSi!Si!S$S$S:S@:S S SISISᐄS@ᐄS_S _SSSSSS SYS YSSd SCS CS!SS !SS]\S ]\SSSSSSSSS!S!SS$SSSlSlSySySqSXqSASASS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g3 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 RY  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g3 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 RY  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g3  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 RY   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`8\),P 10T|9$HA(Lt 0 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1Im#G!jDhBe?c=`: ^   8  =a;^8\6Y}3W b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r0_p64_c0_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements1800input0Elements1800invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1<I[$m6HZQ  c ,  bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  TT)D*D44>4?4IIS T ^^hisjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$ | |""$l%l'')\*\,,.L/L113<4<668,9,;;=>@@B C EEGGItJtLLN`O`QQSLTLVVX8Y8[[]$^$``bceeggirjrlln`o`qqsLtLvvx8y8{{}$~$XXΌΎDD  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xMSʃSʃSSdSSlSc!Sc!S$S$S9S@9S]~S]~SCSCSېS@ېSYS YSqSqSSSS SSS SSSd SCS CSSS SS\S \S5S5SͅSͅS_S_S[S[SY)S Y)SSSSpSlStlSySTySu Su SE߉S$E߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations17k b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ?7 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations17k b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ?7 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations17k  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ?7  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`!Ex_6W{7Fj{%Iz/S8Y} b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@u2@dr /=ao,:^l  )  7 [  i  &  4Xf#1U'Kc8\ Im*N#Z~X|4k!E|AKoy4>bl'1U_  $!H!!!R"v"" ###$;$$$E%i%%& b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1duj_TI>3(   x M   k ;  ^ r  Q )D7* bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  II)/*/44>?IISS]n^nhhrPsP} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``LL 8 8  $$llR R""$8%8'')*,,./1w1w335]6]88:C;C==? @ B}B}DDF_G_IIK\L\NNP>Q>SSU V XXZ[]s]s__a]b]ddf?g?iik!l!nnpqststuuwVxVzz|8}8ll܋܍LL  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xMS5˃S5˃SSdSSlSi!Si!S$S$S:S@:SSSISISᐄS@ᐄS_S _S%S%SSSS SYS YSSd SCS CSORS ORS[S [SӚSӚS…S…SmSmSSS%S%SS$SSSlSlSySyS SX S߉S߉SqS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i5 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 SY  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i5 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 SY  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i5  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 SY   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`<p[:PtY}0T8W{4)Ml6Z b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations17[Hl"Y}3j D{1U BfSw -d 1z Tx.R,uOs)M' b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r0_p64_c0_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements1800input0Elements1800invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1<7n%\J8o& u > P bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  NN):*:44>&?&IIST^^hhrSsS} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&tt` `""$L%L'')8*8,,.$/$11346688:r;r==?^@^BBDJEJGGI6J6LLNOQQSTVzVzXXZb[b]]_J`Jbbd2e2ggijllnoqzqzssubvbxxzJ{J}}22ss[[ώϐCC  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSʃSʃSSdSSlSu!Su!S$S$S9S@9Sc~Sc~SUSUS퐄S@퐄SkS kSwSwSSSS SeS eSSd S+CS +CSQS QS[S [SISIS帅S帅SSSSS S SS$SSSlSlSySyS SX S߉S߉SqS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations18l b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 7 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations18l b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 7 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations18l  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 7  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`!E_Vz+O9Jn -]>b!*N2Vw b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1cq*8"F0T>bLpZ ~   h   v  ! -  f  / D } 7[Nr,eC b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r0_p64_c0_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements1800input0Elements1800invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1<cF) oR5{F^8 q *  bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  NN)8*844>#?#IIS T ^^hhrOsO} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaQQ A A  11!!m m""$W%W'')A*A,,.+/+11346688:t;t==?^@^BBDHEHGGI3J3LLNOQQSTVwVwXXZ_[_]]_G`Gbbd/e/ggijllnnpsqsssu[v[xxzC{C}}++ooYYΎΐCC  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lSS)˃S)˃SSLSSTS{!S{!S$S$S:S(:SSS[S[SS(SqSqSSS}S}SS SkS kSSL S1CS 1CSsRS sRS[S [SSSSSSSSDSSS/lS /lSSSSS#S0#SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\( b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ] bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\( b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ] bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\(  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ]  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`-Q|n$Quq$Heu+Y}Hl+OW{ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ W{-Q'nDh>[ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1g. u <   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  RR)@*@44>.?.IIST^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddRR @ @  ..  o o""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGIVJVLLN@O@QQS*T*VVXY[[]]_s`sbbd]e]ggiGjGlln1o1qqstvvxy{z{z}}ddقلNNÇÉ88"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSʃSʃSSdSSlSu!Su!S$S$S9S@9Sc~Sc~SUSUS퐄S@퐄SkS kSwSwSSSS SeS eSSd S+CS +CSRS RS[S [S'S'SSS_S_SSS$S$SᩆS$ᩆSSSkSkSmySmyS SX S߉S߉S/S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations18l b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 7 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations18l b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 7 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations18l  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 7  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`>h=2V;_ 0Jn$H_b<` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1<` \@|<` \@|<` \@| < ` |6ZMq+dB{5Y b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r0_p64_c0_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements1800input0Elements1800invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1<<x4p,h$`X c  U  G bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  LL)6*644> ? IIST^w^whhrBsB} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%qq[ [""$E%E'')/*/,,./11346x6x88:b;b==?L@LBBD6E6GGI J LLNOQyQySSU_V_XXZE[E]]_+`+bbdeggiikelennpIqIssu-v-xxz{}}ggلن22  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lSS/˃S/˃SSLSSTS{!S{!S$S$S:S(:SSS[S[SS(SqSqSSS}S}SS SkS kSSL S1CS 1CSQS QSZS ZS7S7S9…S9…SSS SD SSSlS lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^* b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ^ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^* b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ^ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^*  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ^  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`,Dhkp Ae$HPt;"Fn2VZ~ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@1U+rHlB_5Y /v bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 J  Y g } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  OO)9*944>#?#IIS T ^|^|hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''uua a""$L%L'')6*6,,. / 113 4 6688:i;i==?S@SBBD=E=GGI'J'LLN O QQSSUfVfXXZL[L]]_2`2bbdeggiikqlqnnpWqWssu=v=xxz#{#}} ||bbՉՋHH.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLSSTS!S!S%S%S:S(:Si~Si~SgSgSS(S}S}S}S}SSSS SwS wSSL S=CS =CSQS QS7[S 7[SSS9S9S S SSDSSS/lS /lSGSGSS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19m b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 59  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19m b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 59  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19m  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 59   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`8<`w:m\ 7[d_5Z~9] b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@  A e # ^ @{9]Vz8s bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1P F  <  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  OO)=*=44>-?-IIST^^hhrqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaOO = =  ++~~l l""$Z%Z'')H*H,,.6/6113$4$6689;;=>@w@wBBDeEeGGIUJULLNCOCQQS1T1VVXY[[] ^ ``bbdrerggi`j`llnIoIqqs-t-vvxy{{}}ggقلKK// b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS#˃S#˃SSLSSTS!S!S%S%S:S(:S~S~SsSsS S( SSSSSwSwSS SS SSL SICS ICSQS QS[S [SOSOS%S%SSSSDSSS/lS /lS5S5SS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~Q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 hY. bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~Q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 hY. bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~Q  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 hY.  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`=a"FX|#>`Pt$H7Im /Ae b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@4 X *oAe7| Nr D[ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1F P  Z  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  XX)D*D44>.?.IIST^^hhrrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hh\\ P P  DD88((  "#%v%v'')b*b,,.N/N113:4:668&9&;;=>@@BBDnEnGGIVJVLLN>O>QQS&T&VVXY[[]]_j`jbbdReRggi:j:lln"o"qqs t v~v~xxzf{f}}NN‚„66vv b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLSSTS!S!S$S$S:S(:Si~Si~SaSaSS(SwSwS}S}SSSS SqS qSSL S7CS 7CSQS QS[S [SOSOS-S-SSS SD SSSlS lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19m b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 48 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19m b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 48 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19m  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 48  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`5cVz)=aJn7>bs2VW{g b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@1UPt2m+O Hl*e bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 w 2 m ( c } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  KK)5*544>!?!IIS T ^}^}hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaKK 5 5    ~~hhR R""$<%<'')&*&,,./11335o6o88:Y;Y==?C@CBBD-E-GGIJLLNOQxQxSSUbVbXXZL[L]]_6`6bbd e ggi j llnnpcqcssuGvGxxz+{+}}ee׉׋II-- b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS)˃S)˃SSLSSTS!S!S%S%S:S(:S~S~SsSsS S( SSSSSwSwSS SS SSL SICS ICSRS RS=[S =[SSSSS}S}SSDSSSlS lS S S߉S0߉S5S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 iY0 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 iY0 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 iY0  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`^%6Z\^!Ef^Aev^\ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@-Q#h:^ 0uGk= bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1  d ) n3 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  RR):*:44> ? IIST^v^vhhrZsZ} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&&n n""$V%V'')>*>,,.&/&11346688:j;j==?P@PBBD4E4GGIJLLNNPnQnSSURVRXXZ6[6]]_`bbddfpgpiikTlTnnp8q8ssuvxxz{}r}rVVȄȆ:: b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLSSTS!S!S)%S)%S :S( :So~So~SSS#S(#SSSSSSSS SS SSL SaCS aCSQS QSZS ZSCSCSSSuSuS#SD#SSS5lS 5lSYSYS)S0)SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1:n b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  O@ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1:n b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  O@ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1:n  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  O@  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`++Otr %[JnnAe DDh'K> b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ ( L `0T$h8\,p bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1X ` $  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  YY)E*E44>1?1IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeYY M M  AA55))  "#%%'')s*s,,._/_113K4K668797;;=#>#@@BCEEGGIqJqLLNYOYQQSATAVVX)Y)[[]^``bbdmemggiUjUlln=o=qqs%t%vvx y {{}}ii݂݄QQŇʼn99!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS˃S˃SSLSSTS!S!S %S %S:S(:S~S~SmSmSS(SSS S SqSqSS S}S }SSL SCCS CCSwQS wQSZS ZSSSSSSS/SD/SSSAlS AlSSS_S0_SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1sF b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 h " bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1sF b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 h " bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1sF  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 h "  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`7[;Bf9!(Lp2b2Vm /@dSw b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ N r @QuCTx"F bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1%h+ n 1  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  ZZ)L*L44>>?>IIS*T*^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhZZ L L  >>00""  "#%%'')q*q,,.c/c113U4U668G9G;;=9>9@@B+C+EEGHJJL M OOQQSrTrVVX`Y`[[]N^N``b:c:eeg$h$jjlmooqqsgtgvvxMyM{{}3~3qqUUǑǓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLS ST S!S!S#%S#%S :S( :So~So~SSSS(SSSSSSSS SS SSL S[CS [CSQS QSZS ZSSS%S%SGSGS SD SSSlS lS S S߉S0߉S_S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1:n b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  N> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1:n b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  N> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1:n  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  N>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`,dDhp 0T/S[ =aEi'+O/Sx b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@d4xHl< Pt DX| bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1h , p 4x } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  UU)=*=44>%?%IIS T ^{^{hhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%% } }""$i%i'')Q*Q,,.9/9113!4!668 9 ;};}==?e@eBBDMEMGGI5J5LLNOQQSSUoVoXXZS[S]]_7`7bbdeggiikqlqnnpUqUssu9v9xxz{}}ssWWɉɋ;; b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS˃S˃SSLSSTS!S!S %S %S:S(:S~S~SmSmSS(SSS S SqSqSS S}S }SSL SCCS CCSwQS wQSZS ZSSS+S+SSSSSSDSSS/lS /lSSS߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1tH b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 i $ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1tH b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 i $ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1tH  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 i $  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`7[&0Te;X|;_i#G:Dh Sw b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@W{%IZ~(L]+O` bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1t 7 z = } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  TT)B*B44>0?0IIST^^hhrlsl} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffTT B B  00  q q""$_%_'')M*M,,.;/;113)4)6689;;=>@|@|BBDjEjGGIXJXLLNBOBQQS,T,VVXY[[]^`u`ubbd[e[ggiAjAlln%o%qqstvxvxxxzZ{Z}}<<rrTT b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTSu!Su!S$S$S:S(:Su~Su~SUSUS퐄S(퐄SkSkSSSSSS SeS eSSL S+CS +CSSQS SQSZS ZSיSיSSSSSSDSSS#lS #lS S S߉S0߉SYS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 *7  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 *7  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 *7   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$:^$Ho&eMqq;_+OOs6 --Q b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ D*y_Ei+O5k bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1Q o >  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  MM)7*744>?IIST^x^xhhr^s^} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&rr\ \""$F%F'')0*0,,./11346y6y88:c;c==?K@KBBD1E1GGIJLLNNPpQpSSUVVVXXZ<[<]]_"`"bbdeg{g{iikalannpGqGssu-v-xxz{}}ll߄߆RRĉċ66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S˃S˃S{SL{SSTSu!Su!S$S$S:S(:S~S~SUSUS퐄S(퐄SkSkSSSkSkSS SeS eSSL S+CS +CSkQS kQSZS ZSSSSSSS;SD;SSSMlS MlSSSqS0qSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h; b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Q] bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h; b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Q] bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h;  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Q]  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`7[ 00Tn%3f3W0T9bDh b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ /vLp"Fc9]3z  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1Z! h / v  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  XX)H*H44>8?8IIS$T$^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhXX H H  88((  ""$p%p'')`*`,,.P/P113@4@668090;;= > @@BCEEGHJxJxLLNdOdQQSPTPVVX?IISS]j^jhhrLsL} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((ppV V""$<%<'')"*",,./1{1{335a6a88:G;G==?+@+BBD E G~G~IIK`L`NNPBQBSSU$V$XXZ[]w]w__aYbYddf;g;iiklnnpprpspuuwRxRzz|4}4hh؋؍HH b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S˃S˃S{SL{SSTSu!Su!S$S$S:S(:S~S~SUSUS퐄S(퐄SkSkSSSkSkSS SeS eSSL S+CS +CS/QS /QSkZS kZSSSc…Sc…SSS#SD#SSS5lS 5lS)S)S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i=  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 P^ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i=  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 P^ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i=   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 P^  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`7[&)M8Z~;__$H,PPt . b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Pt&J g=a7~ Tx*N bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 =  K Y } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  RR)>*>44>*?*IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **xxd d""$P%P'')<*<,,.(/(11346689;v;v==?b@bBBDNENGGI:J:LLN"O"QQS T V~V~XXZf[f]]_N`Nbbd6e6ggijllnoqzqzssubvbxxzJ{J}}22vv^^ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSSLSSTS!S!S$S$S:S(:S{~S{~SaSaSS(SwSwSSSSSS SqS qSSL S7CS 7CSQS QSZS ZS+S+SSSSS#SD#SSS5lS 5lS/S/S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1<p b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 8  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1<p b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 8  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1<p  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 8   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`7[&&J|C>p+2{d&cX|| b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@"|xtplh d bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1pJ $ ~ X  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  RR)<*<44>'?'IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%%q q""$[%['')E*E,,.///11346689;x;x==?b@bBBDLELGGI7J7LLNOQQSTV{V{XXZc[c]]_K`Kbbd3e3ggijllnoqwqwssu_v_xxzG{G}}//wwaa b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS˃S˃SuSLuSSTS{!S{!S$S$S:S(:S~S~S[S[SS(SqSqSSS_S_SS SkS kSSL S1CS 1CS5QS 5QSqZS qZSSSSSkSkS/SD/SSSAlS AlSkSkS;S0;SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]0 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 T  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]0 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 T  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]0  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 T   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`7[&&Jq(Y}Oss#G-QSw  / b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$vbNr:^&J6"t bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1H l >   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  VV)D*D44>2?2IIST^^his{s{} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhVV D D  22  s s""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIZJZLLNDODQQS.T.VVXY[[]^`w`wbbdaeaggiKjKlln5o5qqstvvx y {~{~}}hh݂݄RRLJlj<<&& b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSSLSSTS!S!S$S$S:S(:Su~Su~SaSaSS(SwSwSSSSSS SqS qSSL S7CS 7CSQS QSZS ZS+S+SƅSƅSSSSDSSS)lS )lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1<p b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 8 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1<p b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 8 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1<p  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 8  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`7[&,Pu, Bt[4KoBf!E'(L| b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@`\X|TxPtLpHlDh bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 2 f@ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  PP):*:44>$?$IIS T ^{^{hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))uu_ _""$I%I'')3*3,,./11346|6|88:f;f==?P@PBBD:E:GGI$J$LLN O Q}Q}SSUcVcXXZI[I]]_/`/bbdeggiikilinnpMqMssu1v1xxz{}}kk݄݆QQĉċ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS˃S˃SuSLuSSTS{!S{!S$S$S:S(:S~S~S[S[SS(SqSqSSSeSeSS SkS kSSL S1CS 1CSYQS YQSZS ZSݙSݙS!ŅS!ŅSSSSDSSS/lS /lS S S߉S0߉SwS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^2 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 S  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^2 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 S  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^2  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 S   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`7[&&Jw4\a&J*NNrr ,P b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@`Lp8\$H4 r ^Jn bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1b 4 X* } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  SS)=*=44>'?'IIS T ^^hhrfsf} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++yye e""$P%P''):*:,,.$/$11346688:m;m==?W@WBBDAEAGGI+J+LLNOQQSSUjVjXXZP[P]]_6`6bbdeggijlulunnp[q[ssuAvAxxz'{'}} ffىًLL22 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS SL SSTS!S!S%S%S:S(:S{~S{~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CSQS QS1[S 1[SySySSSOSOS/SD/SSSAlS AlSYSYS)S0)SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 7 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 7 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 7  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`7[&&J, 00TsIm::^C DDh b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@i{(:(L:^Lp^  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1^ C ( r  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>1?1IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIYJYLLNGOGQQS5T5VVX#Y#[[]^``bbdvevggidjdllnMoMqqs1t1vvxy{{}}kk݂݄OO33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS ˃S ˃SoSLoSSTS!S!S$S$S}:S(}:S~S~SaSaSS(SwSwSSSYSYSS SqS qSSL S7CS 7CSQS QSZS ZSSSɿSɿSSS/SD/SSSAlS AlSMSMSS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R% b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Wh bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R% b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Wh bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R%  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Wh  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$HowEi6e?$Sww6\Ae b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@`bd f h jln bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations16  p M *  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  \\)H*H44>2?2IIST^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ll`` T T  HH<<,,  "#%z%z'')f*f,,.R/R113>4>668*9*;;=>@@BBDrErGGIZJZLLNBOBQQS*T*VVXY[[]]_n`nbbdVeVggi>j>lln&o&qqstvvxxzj{j}}RRƂƄ::!!zz b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSSLSSTS!S!S%S%S:S(:S{~S{~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CSQS QS[S [SISISMɅSMɅSSSSDSSS/lS /lSSS߉S0߉S}S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 7 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 7 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 7  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`7[&&J .""Fu|KoBx!E>t b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@p /A/SAeSw e !w!! bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1W <!k } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  OO)9*944>%?%IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeOO 9 9  ##  llV V""$@%@'')***,,./11335s6s88:];]==?G@GBBD1E1GGIJLLNOQ|Q|SSUfVfXXZP[P]]_:`:bbd$e$ggijllnnpgqgssuKvKxxz/{/}}iiۉۋMM11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS ˃S ˃SoSLoSSTS!S!S$S$S:S(:S~S~SaSaSS(SwSwSSS_S_SS SqS qSSL S7CS 7CSQS QSZS ZS1S1SDžSDžSWSWSSDSSS)lS )lS S S߉S0߉SMS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S' b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 V i bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S' b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 V i bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S'  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 V i  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$Hx~  Tx Dz0T@Eiv 0<zKo b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@prtvxz|!~#  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 dA{ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  VV)>*>44>$?$IIST^z^zhhr^s^} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjZZ J J  ::**r r""$Z%Z'')B*B,,.*/*11346688:n;n==?T@TBBD8E8GGIJLLNOQrQrSSUVVVXXZ:[:]]_`bbdegtgtiikXlXnnp>  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS˃S˃SuSuSS$S:S:S~S~S[S$[SSSeSeSS$SYQSYQSZS4ZSST S_ S _ SMS MS%S %S_lS _lSS SSSSSS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1AuO- b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1AuO- b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1AuO- b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Dh8|Lp@  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10zz  nn)Z*Z44>D?DIIS,T,^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1zznn b b  VVJJ>>22  "&#&%%'(**,,.t/t113`4`668L9L;;=8>8@@B$C$EEGHJJLLNlOlQQSTTTVVXI?IIIS5T5^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1uugg Y Y  KK==//!!  "#%%'(*~*~,,.p/p113b4b668T9T;;=F>F@@B8C8EEG*H*JJLMOOQRT}T}VVXkYk[[]Y^Y``bEcEeeg/h/jjlmooqqsrtrvvxXyX{{}>~>$$zz^^БГ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS˃S˃S{S{SS$S:S:S~S~S[S$[SSSkSkSS$SSQSSQSZS4ZSST S_ S _ S5S 5S S  SGlS GlS S  S߉S߉SS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1BvM0 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1BvM0 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1BvM0 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ?c%g?c9  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10xx  hh)P*P44>6?6IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1xxhh X X  HH88((  "#%|%|'')d*d,,.L/L1134446689;;=>@x@xBBD`E`GGIFJFLLN*O*QQSTVVXXZd[d]]_H`Hbbd,e,ggijllnnpfqfssuJvJxxz.{.}}hhډڋLL00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS˃S˃SiS iSSS}:S}:S~S~SgSgSSpSYSYSSSQS`QSZSZSS$ S_ S _ S;S ;SS SMlS MlS;S ;S S SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 rHlB_  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10uu  cc)Q*Q44>=?=IIS%T%^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1uucc Q Q  ??--   ""$n%n'')\*\,,.J/J113848668&9&;;=>@@BCEyEyGGIeJeLLNOOOQQS9T9VVX#Y#[[] ^ ``bbdhehggiNjNlln2o2qqstvvxxzg{g}}II++ }}__ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSS SSS/:S/:S~S~SsSsSSpS S SSSeQS`eQSZSZSS$ SY S Y S)S )SS S;lS ;lS S  S߉S߉SwS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 l6Z$e/S  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  ^^)H*H44>.?.IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ss__ K K  77##m m""$W%W'')A*A,,.+/+11346688:t;t==?Z@ZBBD@E@GGI&J&LLN O QQSSUeVeXXZK[K]]_1`1bbdeggiikplpnnpVqVssu???IIS+T+^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qqaa Q Q  AA11!!  "#%y%y'')i*i,,.Y/Y113I4I668999;;=)>)@@BCEEG H JJLLNkOkQQSWTWVVXCYC[[]/^/``bceeghj}j}llnioiqqsUtUvvxAyA{{}-~-{{ggݑݓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSS SSS):S):S~S~SsSsSSpSSSSSeQS`eQSZSZSS$ SY S Y SS SS S)lS )lS S  Si߉Si߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 m7[%f0T  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  \\)B*B44>$?$IIST^w^whhrWsW} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ss__ K K  77##  e e""$K%K'')1*1,,./11335p6p88:V;V==?8@8BBDEGGIIKmLmNNPOQOSSU1V1XXZ[]]__afbfddfHgHiik*l*nnp q s}s}uuw_x_zz|A}A##ssSSÐÒ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS ˃S ˃SoS oSSS}:S}:S~S~S=S=SSpSYSYSSSQS`QSZSZSS$ SY S Y S5S 5S S  SGlS GlS;S ;S S SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N" b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1i} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N" b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1i} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N" b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1i k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 uImC`  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  ]])I*I44>3?3IIST^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qq]] I I  55!!  o o""$[%['')G*G,,.3/31134668 9 ;;==?m@mBBDYEYGGICJCLLN+O+QQSTVVXXZo[o]]_W`Wbbd?e?ggi'j'llnoqqssukvkxxzS{S}};;## gg b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃSS SSS#:S#:S~S~S[S[SSpSSSSS_QS`_QSZSZSS$ SY S Y S/S /SS SAlS AlS;S ;S S SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 f0T_)M  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10rr  __)I*I44>2?2IIST^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1rrbb R R  BB22  ~ ~""$h%h'')R*R,,.e>ggi&j&llnoqqssujvjxxzR{R}}::"" jj b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS˃S˃SuS uSSS:S:S~S~SgSgSSpS_S_SSSQS`QSZSZSS$ SY S Y S;S ;SS SMlS MlSqS qSASASS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X+ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X+ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X+ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 kAe<Y}  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  [[)I*I44>5?5IIST^^h i s~s~} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mm[[ I I  77%% x x""$f%f'')T*T,,.B/B1130406689;;= > @@BBDqEqGGI]J]LLNGOGQQS1T1VVXY[[]^`z`zbbddedggiNjNlln8o8qqs"t"vvx y {{}}kkUUʇʉ??)) b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃSS SSS#:S#:S~S~S[S[SSpSSSSSQS`QSZSZSS$ SY S Y S#S #SS S5lS 5lS S  S߉S߉SkS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 v X|:^@  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  [[)E*E44>-?-IIST^^hhrhsh} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp\\ H H  44    j j""$T%T'')>*>,,.(/(11346688:q;q==?[@[BBDEEEGGI-J-LLNOQQSSUlVlXXZR[R]]_8`8bbdeggijlrlrnnpVqVssu:v:xxz{}}ttXXˉˋ>>$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS˃S˃SuS uSSS:S:S~S~SgSgSSpS_S_SSSQS`QSZSZSS$ SY S Y S#S #SS S5lS 5lS S  S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W+ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W+ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W+ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 tHlB_  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  ZZ)D*D44>,?,IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nnZZ F F  22  l l""$W%W'')A*A,,.+/+11346688:t;t==?^@^BBDHEHGGI0J0LLNOQQSSUoVoXXZU[U]]_;`;bbd!e!ggijlzlznnp`q`ssuFvFxxz,{,}}kkމދQQĎĐ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃSS S S S:S:S~S~SSSSpSSSSSYQS`YQSZSZSS$ SS S S S5S 5S S  SGlS GlS_S _S/S/SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations16} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations16} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations16 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 f0T_)M  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  \\)J*J44>8?8IIS&T&^^h i s|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nn\\ J J  88&& y y""$g%g'')U*U,,.C/C1131416689;;= > @@BBDrErGGI`J`LLNNONQQS1?1IIST^^hissss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mmaa U U  II==--  "#%{%{'')g*g,,.S/S113?4?668+9+;;=>@@BBDqEqGGIYJYLLNAOAQQS)T)VVXY[[]]_m`mbbdUeUggi=j=lln%o%qqs t vvxxzi{i}}QQłń99ww b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS S  S S S:S:S~S~SSSSpSSSSSQS`QS[S[SS$ SS S S S#S #SS S5lS 5lSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations17} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations17} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations17 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 u [Ae'K  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  VV)@*@44>*?*IIST^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1llVV @ @  **ss] ]""$G%G'')1*1,,./11346z6z88:d;d==?N@NBBD8E8GGI"J"LLN O QQSSUkVkXXZU[U]]_?`?bbd)e)ggijllnnplqlssuPvPxxz4{4}}nnRRĎĐ66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS˃S˃SuS uSSS:S:S~S~S%S%SSpSeSeSSSQS`QSMZSMZSS$ SS S S SS SS S#lS #lS S  S߉S߉SGS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`4 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Y} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`4 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Y} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`4 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Y k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 7[Lp:{  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  YY)A*A44>%?%IIS T ^{^{hhr]s]} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mm]] M M  ==-- u u""$]%]'')E*E,,.-/-11346688:q;q==?U@UBBD9E9GGIJLLNOQsQsSSUWVWXXZ;[;]]_`bbdeguguiikYlYnnp=q=ssu!v!xxz{}w}w[[̈́͆== b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃSS SSS:S:S{~S{~S[S[SSpSSSSSkQS`kQSZSZSS$ SM S M S/S /SS SAlS AlSeS eS5S5SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 h8\,p@d  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  bb)N*N44>8?8IIS T ^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nnbb V V  JJ>>22&&  "#%%'(*|*|,,.h/h113T4T668@9@;;=,>,@@BCEEGHJxJxLLN`O`QQSHTHVVX0Y0[[]^``bcetetggi\j\llnDoDqqs,t,vvxy{{}}ppXẊ̉@@(( b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S˃S˃S{S {SSS:S:S~S~S[S[SSpSkSkSSSAQS`AQS}ZS}ZSS$ SM S M S5S 5S S  SGlS GlSS S_S_SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j=  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j=  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j=  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Cg5xFj9  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  [[)M*M44>=?=IIS)T)^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ii[[ M M  ??11##  "#%%'')r*r,,.d/d113V4V668H9H;;=:>:@@B,C,EEGHJJL M OOQQSqTqVVX_Y_[[]M^M``b9c9eeg#h#jjl m ooqqsftfvvxLyL{{}2~2nnRRđē b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃSS SSS:S:Su~Su~S[S[SSpSSSSSQS`QS[S[SS$ SM S M SS SS S)lS )lS S  S߉S߉SeS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 v`Jn4X  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  \\)D*D44>*?*IIST^^hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ll\\ L L  <<,,   ""$p%p'')X*X,,.@/@113(4(6689;;==?l@lBBDTETGGI:J:LLNOQQSTVtVtXXZX[X]]_<`v>xxz"{"}}xx\\Ή΋@@$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S˃S˃S{S {SSS:S:S~S~S[S[SSpSkSkSSS#QS`#QS_ZS_ZSS$ SM S M SS SS S/lS /lSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i=  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i=  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i=  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 7["c-Q  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  WW)E*E44>1?1IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiWW E E  33!!t t""$b%b'')P*P,,.>/>113,4,6689;;=>@@BBDmEmGGIYJYLLNCOCQQS-T-VVXY[[]^`v`vbbd\e\ggiBjBlln&o&qqstvyvyxxz[{[}}==qqSS b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYSʃSʃSS SSS:S:Su~Su~SySySSpSSSSS=RS`=RSy[Sy[SS$ SM S M S S  SS SlS lS S  S߉S߉SYS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations13} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations13} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations13 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 es,:  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  RR)<*<44>"?"IIST^{^{hhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++wwa a""$K%K'')5*5,,./113 4 6~6~88:h;h==?N@NBBD4E4GGIJLLNOQsQsSSUYVYXXZ?[?]]_%`%bbd e g~g~iikdldnnpJqJssu0v0xxz{}}ooSSʼnŋ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYS˃S˃SS SSS:S:S~S~SSS Sp SqSqSSS#QS`#QS_ZS_ZSS$ SM S M S/S /SS SAlS AlSS SeSeS S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1sF b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Y} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1sF b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Y} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1sF b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Y k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Bf .n6Z"  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  UU)E*E44>3?3IIST^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%% } }""$m%m'')]*],,.M/M113=4=668-9-;;=>@@B C EEGGIsJsLLN_O_QQSKTKVVX7Y7[[]#^#``bceeggiqjqlln]o]qqsItIvvx5y5{{}!~! oo[[ёѓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYSʃSʃSS SSS :S :So~So~SySySSpSSSSSQS`QSZSZSS$ SM S M SS SS S lS  lS{ S { SK߉SK߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations14} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations14} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations14 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Nr2V:o  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  PP)6*644>?IISS]k^khhrKsK} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++ssY Y""$?%?'')%*%,,. / 1~1~335d6d88:J;J==?,@,BBDEGGIIKaLaNNPCQCSSU%V%XXZ[]x]x__aZbZddf'?'IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))wwc c""$O%O'');*;,,.'/'11346688:u;u==?a@aBBDMEMGGI7J7LLNOQQSTV{V{XXZc[c]]_K`Kbbd3e3ggijllnoqwqwssu_v_xxzG{G}}//ss[[ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYSʃSʃSS SSS :S :So~So~SOSOSSpSSSSS RS` RSI[SI[SS$ SG S G SS SS S)lS )lS#S #S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ^bf j  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  SS)=*=44>&?&IIST^^hhrlsl} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&&r r""$\%\'')F*F,,.0/011346689;y;y==?c@cBBDMEMGGI6J6LLNOQQSTVzVzXXZb[b]]_J`Jbbd4e4ggijllnoqxqxssu`v`xxzH{H}}00vv`` b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYS#˃S#˃SS SSS:S:S~S~SOSOSSpSwSwSSSPS`PS;ZS;ZSS SG S| G S)S )SQ~S Q~S/NS /NS_S _SUމSUމSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|O b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1Ei; s  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  OO)=*=44>)?)IIST^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaOO = =  ++~~l l""$Z%Z'')H*H,,.6/6113$4$6689;;=>@w@wBBDeEeGGIQJQLLN;O;QQS%T%VVXY[[]]_n`nbbdXeXggiBjBlln6o6qqs t vvx y {{}}iiނބSSȇȉ=='' b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYSʃSʃSS SSS:S:Si~Si~SOSOS}Sp}SSSSS5QS`5QSqZSqZSS$ SG S G S S  SS SlS lS S  S߉S߉SSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Cg 1r=a7  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  OO)9*944>!?!IIS T ^|^|hhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((tt^ ^""$H%H'')2*2,,./11346{6{88:e;e==?O@OBBD9E9GGI!J!LLNOQzQzSSU`V`XXZF[F]]_,`,bbdeggiikjljnnpNqNssu2v2xxz{}}llބކPPÉË66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYS#˃S#˃SS SSS:S:S~S~SOSOSSpSwSwSSSSS`SS\S\SS SG S| G SS S9~S 9~SNS NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{O b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations17[Quk  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  NN)8*844> ? IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&tt` `""$K%K'')5*5,,./113 4 6~6~88:h;h==?R@RBBD.?.IIST^^h i s~s~} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGIVJVLLNDODQQS2T2VVX Y [[]^``bbdsesggiajallnVoVqqs:t:vvxy{{}~ttXXʇʉ<<  b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LSS)˃S)˃SS SSS:S:S~S~SSSSpSwSwSSSaRS`aRS[S[SS SA S| A S#S #SK~S K~S)NS )NSAS AS7މS7މSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X$ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Y} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X$ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Y} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X$ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Y k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations18\p 0Dh  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  QQ)G*G44>/?/IIS#T#^^h i s}s}} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaUU I I  ==11!!   ""$y%y'')e*e,,.Q/Q113=4=668)9);;=>@@BBDoEoGGIWJWLLN?O?QQS'T'VVXY[[]]_k`kbbd_e_ggiGjGlln/o/qqstvvxxzs{s}}[[ςτCC(( b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LSSʃSʃSS SSS9S9Sc~Sc~SmSmSwSpwSSSSSRS`RS=[S=[SS SA S| A SS S?~S ?~SNS NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations18l b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations11} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations18l b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations11} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations18l b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations11 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations17[ W{,P  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  JJ)8*844>"?"IIS T ^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``JJ 4 4  }}ggQ Q""$?%?''))*),,./11335r6r88:\;\==?F@FBBD0E0GGIJLLNOQyQySSUcVcXXZM[M]]_7`7bbd!e!ggi j llnnppqpssuTvTxxz8{8}}rrVVȎȐ:: b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LSS)˃S)˃SS SSS:S:S~S~SSSSpSwSwSSSSQS`SQSZSZSS SA S| A S S  S3~S 3~SNS NS S  S݉S݉SgS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X$ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Y} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X$ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Y} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X$ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Y k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1u_Im3W  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  MM)=*=44>!?!IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaQQ A A  11!!  i i""$Y%Y'')A*A,,.)/)11346688:m;m==?Q@QBBD5E5GGIJLLNNPoQoSSUSVSXXZ7[7]]_`bbd e g}g}iikalannpEqEssu)v)xxz { }}ccՄՆEE''  b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations1MScʃScʃSS|SmSmSSS"S|"S&SH&S9S9S?~SP?~S僄SX僄ScScSSPSySySSSSSS$ S]S, ]SS SsS$ sSS S9DS| 9DSSSXSS]S|]SeSeS߾Sx߾SS\SSS$S$S5S85S S SGlSGlSySySSlSkSkSS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Pm b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1) bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Pm b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1)  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Pm  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1) bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`<`9]AX|':vu 4XY}Pt2n b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@)M9]Ptd .6Zg,!E~r8\Dh:Pt  .Qug!E~8\ O s   - f   D }  7 [    N r  ,eC|6ZMq+dB{5YLp*cA b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d)9rn+3dx=VH:t-f   X   J   <  u bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10CC  OO)I*I44>9?9IIS%T%^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1CC33 # #  //w w""$q%q'')a*a,,.Q/Q113A4A668191;;=!>!@@BCEEGHJyJyLLNeOeQQSQTQVVX=Y=[[])^)``bceeghjwjwllnoooqqs[t[vvxGyG{{}3~3 mm# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xMS5˃S5˃SSdSSlSW!SW!S$S$S:S@:SSS7S7SϐS@ϐSMS MS%S%SSSS SGS GSSd S CS  CS+RS +RSg[S g[SSS…S…S[S[SSS%S %SSSSpS/lSt/lSySTySMSMSS$SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n: b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 E-~ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n: b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 E-~ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n:  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 E-~  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Bqc,PKoq&J:^',Jnn: b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@1UGkA3~^\4X:^ .<uKoe!ECgb8\ 2 y   O s  % I    f  < `  :]7[5~X|2V 0y Sw-Q+tNr(L&o b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d1G|UGc]*q8FX  ! j 3  |  E  W  bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  PP)@*@44>0?0IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$]%]'')K*K,,.9/9113'4'6689;;==?^@^BBDLELGGIXJXLLNFOFQQS4T4VVX"Y"[[]^``bbdueuggicjcllnXoXqqs>""  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1MScʃScʃSS|SmSmSSS"S|"S&SH&S9S9S9~SP9~S僄SX僄ScScSSPSySySMSMSS$ S]S, ]SS SsS$ sSS S9DS| 9DSURSXURS[S|[SٚSٚSDžSxDžSYS\YSuSuSs'Ss'SS8SSS/lS/lSySyS#Sl#S߉S߉SS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Oj b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1) bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Oj b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1)  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Oj  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1) bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`*N]h_Bf 1a[ c'Ku%?c b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@]_ace g i#Gkm^ 0 T o    & q k   = a s    3 u x    w  y{ {ywusqomki g eca_]  b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d]:tQ.G  hQE"}  X  3   i D zU0 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10BB  KK)?*?44>+?+IIST^^his{s{} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1BB..    ##qq] ]""$Q%Q'')=*=,,.)/)11346689;w;w==?c@cBBDOEOGGI;J;LLN#O#QQS T VVXXZg[g]]_O`Obbd7e7ggijllnoqqssuovoxxzW{W}}??''kk# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xMS;˃S;˃SSdSSlSW!SW!S$S$S:S@:SSS7S7SϐS@ϐSMS MS%S%SSSS SGS GSSd S CS  CSSS SS\S \SASASąSąSSSSS&S &SSSSpS#lSt#lSySTyS S S߉S$߉SkS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p< b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 F. bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p< b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 F. bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p<  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 F.  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`1U6Pt"F ._2V=~Bf8\Gk b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Rv>b*N:&xdPt<`!E#(Ln  8 L p   $ * N v    , b w  N r U y  : ^ 3 W  &J4lTx<`$H 0hPt8\ D,|dLp4X@ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dR$vH(ls>> bT4 T $  t  D   d 4T bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  JJ)8*844>$?$IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``JJ 4 4  }}ggQ Q""$?%?''))*),,./11335r6r88:\;\==?)@)BBDEGGIJLLNOQ{Q{SSUeVeXXZO[O]]_9`9bbd#e#ggi j llnoqrqrssuVvVxxz:{:}}ttXXʎʐ<<  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xMSʃSʃSSdSSlSi!Si!S$S$S9S@9Sc~Sc~SISISᐄS@ᐄS_S _SwSwSSSS SYS YSSd SCS CSSS SS/]S /]SwSwS待S待SSSSS$S $SSSSpS/lSt/lSySTySYSYS)S$)SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19m b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 <9 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19m b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 <9 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19m  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 <9  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`9]?c#\a@ D?c#Hl\@ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@9]"FPtc .9]g!E3~z 8\Pt&JOs - fg D}7[ O s   - f   D }  7 [    N r  ,eC|6ZMq+dB{5YLp*cA b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d9"ri+0dw>VL>w0i  " [   M   ?  x bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  MM);*;44> ? IIST^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__MM ; ;  ))||j j""$X%X'')F*F,,.4/4113"4"6689;;==?u@uBBDcEcGGI5J5LLNOQQS T VVXY[[]]_n`nbbdXeXggiBjBlln6o6qqs t vvx y {{}}iiނބSSȇȉ==''  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xMS/˃S/˃SSdSSlSi!Si!S$S$S:S@:S S SISISᐄS@ᐄS_S _SSSSSS SYS YSSd SCS CSmRS mRS[S [SSS1S1SSSSSSSS$SSS lS lSySyS/SX/S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c/ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 PW  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c/ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 PW  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c/  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 PW   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ce61Uw$&JCg6:^Uy$HNr b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1Hl,P D*sdMq<`)M8'pX|Jn0T$H,tLp$ H   h  b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r0_p64_c0_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements1800input0Elements1800invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1<H,uX> hP0x@P bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  ZZ)F*F44>2?2IIST^^his\s\} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffZZ N N  BB66**  "#%%'')t*t,,.`/`113L4L668898;;=$>$@@BCEEGGIrJrLLNZOZQQSBTBVVX*Y*[[]^``bbdnenggiVjVlln@o@qqs(t(vvxy{{}}llTT ||  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xMSʃSʃSSdSSlSc!Sc!S$S$S9S@9Sc~Sc~SCSCSېS@ېSYS YSwSwSSSS SSS SSSd SCS CSCRS CRS[S [SǚSǚSSSSSSS$S $S S SSpSlStlSySTyS S S߉S$߉SeS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19m b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ;9 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19m b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ;9 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19m  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ;9  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`>b+Oh(?ck|*EiSwC>g.R b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@9]Pt .g!E~8\Os -f D=a}77[~  N r T x   , * N e  $  C | k  6 Z A e     ; M q  +dB}7[Nr .i#G:^Uy3l(L b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d9r+dWLIZ;!t-f   X   J   <  u bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  LL)6*644>?IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``LL 8 8  $$rr^ ^""$I%I'')3*3,,./11346|6|88:f;f==?P@PBBD:E:GGIJLzLzNNP`Q`SSUcVcXXZI[I]]_/`/bbdeggiiknlnnnp\q\ssuBvBxxz({(}}ggډڋMM33  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xMS/˃S/˃SSdSSlSi!Si!S$S$S:S@:S S SISISᐄS@ᐄS_S _SSSSSS SYS YSSd SCS CSSS SS]S ]SYSYS%S%SSSSSSSᩆS$ᩆSSSkSkSmySmyS SX S߉S߉S5S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e1 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 QW  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e1 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 QW  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e1  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 QW   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`2VmGkTx4!Ef"FhDh$Vz b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations19]Pt .g!E~8\Os -f" Dk}7[Ei  C  f @ d  >  a  b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r0_p64_c0_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements1800input0Elements1800invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1<9r+dXbJ+t<= bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  TT)<*<44>$?$IIST^~^~hhrEsE} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$ | |""$h%h'')P*P,,.8/8113 4 6689;|;|==?d@dBBDLELGGI4J4LLNOQQSSUnVnXXZR[R]]_6`6bbdeggiikplpnnpXqXssu)?)IIST^^hhrQsQ} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccWW K K  ??33##  ""$q%q'')]*],,.I/I113545668!9!;;= > @@BBDiEiGGIQJQLLN9O9QQS!T!VVX Y [}[}]]_e`ebbdMeMggi5j5llnoqqstvyvyxxza{a}}II11ooUU  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lSS)˃S)˃SSLSSTS{!S{!S$S$S:S(:SSS[S[SS(SqSqSSS}S}SS SkS kSSL S1CS 1CSQS QS [S  [SUSUS)S)SSSSDSSSlS lS S S߉S0߉SSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X$ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 [ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X$ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 [ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X$  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 [  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Sw<W{=a2Mq'K(Cg59]u b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@C`6Z 0wMq#Gd bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1f-t; } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  NN)8*844> ? IIST^y^yhhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''ss] ]""$G%G'')1*1,,./11346z6z88:d;d==?L@LBBD2E2GGIJLLNNPqQqSSUWVWXXZ=[=]]_#`#bbd e g|g|iikblbnnpHqHssu.v.xxz{}}mmSSʼnŋ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSʃSʃSSdSSlSu!Su!S$S$S:S@:Si~Si~SUSUS퐄S@퐄SkS kS}S}SSSS SeS eSSd S+CS +CSRS RS3\S 3\S{S{SSSSSSSSS۩S$۩SSSkSkSgySgyS SX Si߉Si߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1:n b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 9 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1:n b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 9 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1:n  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 9  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`&J`/S.Rhg6Zp'>bx;_  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1GkA^4X .uKo!EbW{8\1U 2 / y x   O s R v , P  * s  M q  b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r0_p64_c0_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements1800input0Elements1800invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1<GUc*qO8aF* bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  OO)7*744>?IIST^s^shhr:s:} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33##k k""$S%S'');*;,,.#/#113 4 6688:g;g==?M@MBBD1E1GGIJLLNNPkQkSSUOVOXXZ3[3]]_`bbddfmgmiikQlQnnp5q5ssuvxxzz|o}oSSńņmmސޒ  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lSS)˃S)˃SSLSSTS{!S{!S$S$S:S(:SSS[S[SS(SqSqSSS}S}SS SkS kSSL S1CS 1CSiSS iSS\S \S훅S훅S״S״SSSSDSSSlS lSu Su SE߉S0E߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z& b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 \ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z& b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 \ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z&  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 \  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`t+O 08\ <`5YlBf(@dMq b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@:^4{ Q u ' K ! h > b  8  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1IW } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  LL)2*244>?IISS]i^ihhrKsK} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''ooU U""$;%;'')!*!,,./1z1z335`6`88:F;F==?*@*BBD E G}G}IIK_L_NNPAQASSU#V#XXZ[]v]v__aXbXddf:g:iiklnnpprosouuwQxQzz|3}3gg׋׍GG b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLSSTS!S!S%S%S :S( :So~So~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CSRS RS9\S 9\SSSYSYSSS)SD)SSS;lS ;lSSSSS0SSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 1;  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 1;  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 1;   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`/Sl|@<`o .3\r @dOs<` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@4];_=fDh"F bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1[&q< } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  UU)G*G44>9?9IIS%T%^^h i s~s~} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccUU G G  99++  "#%z%z'')l*l,,.^/^113P4P668B9B;;=4>4@@B&C&EEGHJJLMOOQQSmTmVVX[Y[[[]I^I``b5c5eeghjjl m o|o|qqsbtbvvxHyH{{}.~.llތގPP‘“ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS#˃S#˃SSLSSTS!S!S%S%S:S(:S~S~SsSsS S( SSS S SqSqSS SS SSL SICS ICSGQS GQSZS ZS˙S˙SSSSSSDSSS)lS )lS#S#S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|M b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 fW. bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|M b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 fW. bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|M  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 fW.  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`9] 1Eiv#;bHl(L?Sw%I b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@6{MqCZ~,P"g9] bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1Wa& } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  SS)=*=44>(?(IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&&r r""$\%\'')F*F,,.0/011346689;y;y==?c@cBBDMEMGGI8J8LLN O QQSTV|V|XXZd[d]]_L`Lbbd4e4ggijllnoqxqxssu`v`xxzH{H}}00xxbb b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLSSTS!S!S%S%S :S( :So~So~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CSQS QS[S [S[S[SISISSSSDSSS)lS )lSSS߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 0: bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 0: bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 0:  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`=pFj >^W{Lp,PY}$:^-Q b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$oMq+ O - x V z 4 X  6  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1Rh3 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  QQ)?*?44>-?-IIST^^hhrisi} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccQQ ? ?  --  n n""$\%\'')J*J,,.8/8113&4&6689;;=>@y@yBBDgEgGGIUJULLN?O?QQS)T)VVXY[[]]_r`rbbdXeXggi>j>lln"o"qqstvuvuxxzW{W}}99ooQQ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS#˃S#˃SSLSSTS!S!S%S%S:S(:S~S~SySySS(SSSSSwSwSS SS SSL SOCS OCSRS RS\S \S]S]SSSSS SD SSSlS lS S S߉S0߉SYS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 gW0 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 gW0 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}O  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 gW0  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Sw$Hx8=a\'KLp 50T Dh b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ /tFj <  S w % I  `  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1k0u: } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  OO)9*944>#?#IIST^z^zhhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((tt^ ^""$H%H'')2*2,,./11346{6{88:e;e==?O@OBBD9E9GGI#J#LLN O Q|Q|SSUbVbXXZH[H]]_.`.bbdeggiikhlhnnpLqLssu0v0xxz{}}jj܄܆PPÉË66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLSSTS!S!S)%S)%S:S(:Su~Su~SSS#S(#SSSSSSSS SS SSL SaCS aCSORS ORS[S [SӚSӚSwSwS S S/SD/SSSAlS AlSSSeS0eS S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1<p b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  Q@ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1<p b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  Q@ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1<p  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  Q@  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`&Juf:^_z(ImNr3W<b,P b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@+x Z~>b D&sUy7[ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1T!n; } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  SS)C*C44>3?3IIST^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33## { {""$k%k'')[*[,,.K/K113;4;668+9+;;=>@@B C EEGGIsJsLLN_O_QQSKTKVVX7Y7[[]#^#``bceeggiqjqlln]o]qqsItIvvx5y5{{}!~! oo[[ёѓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS˃S˃SSLSSTS!S!S %S %S:S(:S~S~SmSmSS(SSSSSkSkSS S}S }SSL SCCS CCS_QS _QSZS ZS㙅S㙅SۭSۭSSS#SD#SSS5lS 5lSSSSS#S0#SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1qB b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 f" bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1qB b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 f" bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1qB  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 f"  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`9]&3W^7))MnZ~@dd4X6 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@)l:^,o=a /r@d2 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1H N } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  TT)B*B44>2?2IIS T ^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffTT B B  00  q q""$_%_'')M*M,,.;/;113)4)6689;;=>@|@|BBDjEjGGIZJZLLNHOHQQS6T6VVX$Y$[[]^``bcewewggiejellnNoNqqs2t2vvxy{{}}llނބPP‡‰44 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLSSTS!S!S)%S)%S :S( :So~So~SSS#S(#SSSSSSSS SS SSL SaCS aCSGQS GQSZS ZS˙S˙SSSSSSDSSS/lS /lS#S#S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1<p b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  P> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1<p b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  P> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1<p  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  P>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` 0d5Ys,P'Aej3W{8\vA* b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@=l N r 0 T  6  e G k  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1U"o< } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  OO);*;44>'?'IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''uua a""$M%M'')9*9,,.%/%11346688:s;s==?_@_BBDKEKGGI7J7LLNOQQSTV{V{XXZc[c]]_K`Kbbd3e3ggijllnoqwqwssu_v_xxzG{G}}//ss[[ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS˃S˃SSLSSTS!S!S %S %S:S(:S~S~SmSmSS(SSS S SqSqSS S}S }SSL SCCS CCSORS ORS[S [SӚSӚSMSMSSSSDSSS#lS #lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1rD b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 g$ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1rD b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 g$ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1rD  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 g$  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`9]Af1UX|z(BEi>[ .R b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@uCg5x F j  8 {  I m  ; ~  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1TZ` } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  NN)8*844>$?$IIST^^hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddNN 8 8  ""  kkU U""$?%?''))*),,./11335r6r88:\;\==?F@FBBD0E0GGIJLLNOQ{Q{SSUeVeXXZO[O]]_9`9bbd#e#ggi j llnnpfqfssuJvJxxz.{.}}hhډڋLL00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTSu!Su!S$S$S:S(:S{~S{~SUSUS퐄S(퐄SkSkSSSSSS SeS eSSL S+CS +CSRS RS[[S [[SSSSSSS)SD)SSS;lS ;lS_S_S/S0/SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 &9  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 &9  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 &9   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` DnSw=a*N] /4Xrg@d>bq b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$u_Im3WA+|f bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1Q"sD } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  QQ)?*?44>-?-IIST^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccQQ ? ?  --  n n""$\%\'')J*J,,.8/8113&4&6689;;=>@y@yBBDgEgGGIUJULLN?O?QQS)T)VVXY[[]]_r`rbbd\e\ggiFjFlln0o0qqstvvxy{y{y}}cc؂؄MM‡‰77!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S˃S˃SuSLuSSTSu!Su!S$S$S:S(:S~S~SUSUS퐄S(퐄SkSkSSSeSeSS SeS eSSL S+CS +CSAQS AQS}ZS }ZSřSřS5S5SSS/SD/SSSAlS AlSeSeS5S05SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f7 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 O[ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f7 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 O[ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f7  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 O[  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`9]&,Pg#&b1Um)M Cgg= b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$mGk!EhBf@c bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1I[$ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  ^^)J*J44>6?6IIST^^hiszsz} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jj^^ R R  FF::..""  "#%%'(*x*x,,.d/d113P4P668<9<;;=(>(@@BCEEGHJvJvLLN^O^QQSFTFVVX.Y.[[]^``bbdrerggiZjZllnBoBqqs*t*vvxy{{}}nnVVʇʉ>>&& b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTS{!S{!S$S$S:S(:Su~Su~S[S[SS(SqSqSSSSSS SkS kSSL S1CS 1CSGQS GQSZS ZS˙S˙SSS-S-SSDSSS)lS )lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 %9  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 %9  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 %9   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$Hef+O2VW{"Uy"Fd? b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Pt:^$ H  2  m  W { A e +O bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1f7Y } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP):*:44>$?$IIS T ^}^}hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((vvb b""$M%M'')7*7,,.!/!113 4 6688:j;j==?T@TBBD>E>GGI(J(LLNOQQSSUgVgXXZM[M]]_3`3bbdeggiikrlrnnpXqXssu>v>xxz${$}} }}cc։֋II// b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S˃S˃S{SL{SSTSu!Su!S$S$S:S(:S~S~SUSUS퐄S(퐄SkSkSSSeSeSS SeS eSSL S+CS +CSRS RS=[S =[SSSSSSSSDSSS)lS )lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g9 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 N\ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g9 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 N\ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g9  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 N\  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`9]:Tx3VzVz@5X|7Z~ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@=a; ^ 8 \  6   Y } 3 W  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1m6H } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  XX)@*@44>(?(IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhXX H H  88((  ""$l%l'')T*T,,.>"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS SL SSTS!S!S$S$S:S(:S{~S{~SaSaSS(SwSwSSSSSS SqS qSSL S7CS 7CSQS QS7[S 7[SSSUSUSISIS)SD)SSS;lS ;lSASASS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 :  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 :  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 :   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`79]:^-Q>@d!E DFj9&JLp b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$$$$$$%% bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1\8p } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  WW)C*C44>-?-IIST^^hhrqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1gg[[ O O  CC77''  ""$u%u'')a*a,,.M/M113949668%9%;;=>@@BBDmEmGGIUJULLN=O=QQS%T%VVX Y [[]]_i`ibbdQeQggi9j9lln!o!qqs t v}v}xxze{e}}MM55uu b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS ˃S ˃SoSLoSSTS{!S{!S$S$S:S(:S~S~S[S[SS(SqSqSSS_S_SS SkS kSSL S1CS 1CSQS QSZS ZS+S+SݮSݮS+S+SSDSSS/lS /lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[, b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 R { bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[, b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 R { bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[,  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 R {  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$Hwn$SwJnn.R&JJn,&&Ju b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$xhX|Hl8\(L<, bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1T(|P } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  RR)<*<44>$?$IIS T ^}^}hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++wwa a""$K%K'')5*5,,./113 4 6~6~88:h;h==?P@PBBD6E6GGIJLLNOQuQuSSU[V[XXZA[A]]_'`'bbd e ggiikflfnnpLqLssu2v2xxz{}}qqWWɉɋ;; b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSSLSSTS!S!S$S$S:S(:S{~S{~SaSaSS(SwSwSSSSSS SqS qSSL S7CS 7CSkQS kQSZS ZSSSYSYSSSSDSSS#lS #lS S S߉S0߉SGS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 : bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 : bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 :  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$:^$Hq'*pSww Jn7[[$H==a b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@% %  %  %  %  % %% bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1L(`< } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  SS);*;44>!?!IIST^w^whhr[s[} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggWW G G  77''o o""$W%W'')?*?,,.'/'11346688:k;k==?Q@QBBD5E5GGIJLLNNPoQoSSUSVSXXZ7[7]]_`bbddfqgqiikUlUnnp9q9ssuvxxz{}s}sWWɄɆ;; b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS˃S˃SuSLuSSTS{!S{!S$S$S:S(:S~S~S[S[SS(SqSqSSS_S_SS SkS kSSL S1CS 1CSQS QS%[S %[SmSmS]S]SKSKSSDSSSlS lS S SW߉S0W߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\. b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Q { bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\. b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Q { bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\.  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Q {  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`3mBf -Gk@d!E'>b!!j<`z b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@p ` P t @ d 0 T D 4 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1$xL t } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  PP)6*644>?IISS]m^mhhrOsO} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++ssY Y""$?%?'')%*%,,. / 1~1~335d6d88:J;J==?.@.BBDEGGIIKcLcNNPEQESSU'V'XXZ [ ]z]z__a\b\ddf>g>iik l nnpqssssuuwUxUzz|7}7kkۋۍKK b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSSLSSTS!S!S%S%S:S(:S~S~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CSQS QS[S [SaSaSSSySyS;SD;SSSMlS MlSSSkS0kSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 9 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 9 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 9  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`9]))M8++O -]-Q9] //S b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$:,PBfX|n3%I bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1gN5 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  YY)K*K44>=?=IIS)T)^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggYY K K  ==//!!  "#%~%~'')p*p,,.b/b113T4T668F9F;;=8>8@@B*C*EEGHJJL M OOQQSqTqVVX_Y_[[]M^M``b9c9eeg#h#jjl m ooqqsftfvvxLyL{{}2~2ppTTƑƓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS˃S˃SiSLiSSTS!S!S%S%S}:S(}:S~S~SgSgSS(S}S}SSSYSYSS SwS wSSL S=CS =CSQS QSZS ZS%S%SSS[S[S/SD/SSSAlS AlS;S;S S0 SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P! b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Uf bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P! b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Uf bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P!  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Uf  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$Hrw6Lpe(L$Sww$mAee b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$* 06<B$H*N bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1_>| } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  WW)A*A44>,?,IIST^^hhrrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjZZ J J  ::** v v""$`%`'')J*J,,.4/411346689;};}==?g@gBBDQEQGGI1?1IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggUU C C  11  r r""$`%`'')N*N,,.@}@}BBDkEkGGIYJYLLNCOCQQS-T-VVXY[[]^`v`vbbd\e\ggiBjBlln&o&qqstvyvyxxz[{[}}==ssUU b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS˃S˃SiSLiSSTS!S!S$S$S}:S(}:S~S~SaSaSS(SwSwSSSYSYSS SqS qSSL S7CS 7CSQS QS[S [S[S[SSSSSSDSSS/lS /lS S S߉S0߉SkS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q# b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 T g bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q# b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 T g bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q#  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 T g  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`,0Ty *Y}.R(9],P>&*N b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@0T6 Z < ` B f H l N r TxZ~ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1[:xW } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  SS)=*=44>'?'IIS T ^~^~hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,xxb b""$L%L'')6*6,,. / 113 4 6688:i;i==?S@SBBD=E=GGI'J'LLN O QQSSUfVfXXZL[L]]_2`2bbdeggiiklllnnpPqPssu4v4xxz{}}nnTTljNj::  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS˃S˃SuSuSS$S:S:S~S~S[S$[SSS_S_SS$SYQSYQSZS4ZSST S_ S _ SYS YS1S 1SklS klSS SSS7S} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=qQ' b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=qQ' b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=qQ' b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ae 0qEiC  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10xx  hh)X*X44>F?FIIS2T2^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1xxhh X X  HH88((  "#%%'')p*p,,.`/`113P4P668@9@;;=0>0@@B C EEGHJJLLNrOrQQS^T^VVXJYJ[[]6^6``b"c"eeghjjllnpopqqs\t\vvxHyH{{}4~4  nn b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃScS cSSSw:Sw:S~S~SaSaS梅Sp梅SSSSSSSQS`QSZSZSS$ S_ S _ SGS GSS SYlS YlSqS qSASASS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 mGk!Ei  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  aa)O*O44>=?=IIS+T+^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ssaa O O  ==++ ~ ~""$l%l'')Z*Z,,.H/H113646668$9$;;=>@@BCEwEwGGIeJeLLNSOSQQSATAVVX/Y/[[]^``b c eeggipjpllnYoYqqs=t=vvx!y!{{}~ww[[͇͉??## b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS˃S˃SuSuSS$S:S:S~S~S[S$[SSSeSeSS$SYQSYQSZS4ZSST S_ S _ SAS ASS SSlS SlSGS GSSSS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>rO* b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>rO* b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>rO* b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 :^"kEiC  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10vv  bb)N*N44>8?8IIS T ^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvbb N N  ::&&t t""$`%`'')L*L,,.8/8113$4$6689;;==?r@rBBD^E^GGIHJHLLN0O0QQSTVVXY[t[t]]_\`\bbdDeDggi,j,llnoqqssupvpxxzX{X}}@@((ll b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃScS cSSSw:Sw:S~S~SaSaS梅Sp梅SSSSSSSQS`QSZSZSS$ S_ S _ S5S 5S S  SGlS GlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 sMq'K%n  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  ]])G*G44>1?1IIST^^hhrqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ss]] G G  11zzd d""$N%N'')8*8,,."/"113 4 6688:k;k==?U@UBBD?E?GGI)J)LLNOQQSSUrVrXXZ\[\]]_F`Fbbd0e0ggijllnoqsqsssuWvWxxz;{;}}uuYYˎː== b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS!S !SSS/:S/:S~S~SsSsSSpS S SSSwQS`wQSZSZSS$ SY S Y SGS GSS SYlS YlS}S }SMSMSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Hl Dd>b  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  bb)P*P44>>,,  ""$m%m'')[*[,,.I/I113747668%9%;;=>@@BCExExGGIdJdLLNNONQQS8T8VVX"Y"[[] ^ ``bbdkekggiUjUlln?o?qqs)t)vvxy{{}}rr\\чщFF00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS˃S˃SiS iSSS}:S}:S~S~S=S=SSpSYSYSSSQS`QSZSZSS$ SY S Y SAS ASS SSlS SlSwS wSGSGSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1f} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1f} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1f k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 mGk!Eh  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  gg)S*S44>=?=IIS%T%^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ssgg [ [  OOCC77++  "#%%' ( **,,.m/m113Y4Y668E9E;;=1>1@@BCEEG H J}J}LLNeOeQQSMTMVVX5Y5[[]^``bceyeyggiajallnIoIqqs1t1vvxy{{}~uu]]чщEE-- b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSS SSS/:S/:S~S~SmSmSSpS S SSSYQS`YQSZSZSS$ SY S Y S/S /SS SAlS AlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ;_$mGk!E  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  __)I*I44>1?1IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ss__ K K  77##q q""$\%\'')F*F,,.0/011346689;y;y==?c@cBBDMEMGGI5J5LLNOQQSTVtVtXXZZ[Z]]_@`@bbd&e&ggi j llnnpeqessuKvKxxz1{1}}ppVVɎɐ<< b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS˃S˃SiS iSSS}:S}:S~S~S=S=SSpSYSYSSSQS`QSZSZSS$ SY S Y S)S )SS S;lS ;lS S  S߉S߉SwS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1g} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1g} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1g k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 tNr(L'p  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  cc)K*K44>1?1IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1sscc S S  CC33##  "#%w%w'')_*_,,.G/G113/4/6689;;==?s@sBBD[E[GGIAJALLN%O%QQS T V{V{XXZ_[_]]_C`Cbbd'e'ggi j l}l}nnpaqassuEvExxz){)}} ccՉՋGG++ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃSS SSS):S):S~S~S[S[SSpSSSSSSQS`SQSZSZSS$ SY S Y S5S 5S S  SGlS GlSSS SS#S#SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 h.RW{A  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  dd)P*P44>8?8IIS T ^^hiszsz} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1tthh \ \  PPDD44   " # %%'')n*n,,.Z/Z113F4F668292;;=>@@BCExExGGI`J`LLNHOHQQS0T0VVXY[[]^`t`tbbd\e\ggiDjDlln,o,qqstvvxxzp{p}}XX̂̄@@%% ~~ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS ˃S ˃SoS oSSS:S:S~S~SaSaSSpS_S_SSSeQS`eQSZSZSS$ SY S Y SS SS S/lS /lS S  S߉S߉SkS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V' b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V' b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V' b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Hl@Z~.R  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  WW)A*A44>'?'IIS T ^^hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1llXX D D  00||f f""$P%P''):*:,,.$/$11346688:m;m==?S@SBBD9E9GGIJLLNOQxQxSSU^V^XXZD[D]]_*`*bbdeggiikilinnpOqOssu5v5xxz{}}ttXXʉʋ<<  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃSS SSS):S):S~S~S[S[SSpSSSSSYQS`YQSZSZSS$ SY S Y SS SS S/lS /lS S  S߉S߉SSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 q7[!`&J  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10rr  ^^)F*F44>*?*IIST^^hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1rrbb R R  BB22 z z""$b%b'')J*J,,.2/211346689;v;v==?Z@ZBBD>E>GGI"J"LLNOQxQxSSU\V\XXZ@[@]]_$`$bbdegzgziik^l^nnpBqBssu&v&xxz { }|}|``҄҆BB$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS ˃S ˃SoS oSSS:S:S~S~SaSaSSpS_S_SSSQS`QSYZSYZSS$ SY S Y S S  SS SlS lS S  Sc߉Sc߉S S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U' b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U' b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U' b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ;_X|?   b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  WW)=*=44>?IIST^r^rhhrRsR} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nnZZ F F  22zz` `""$F%F''),*,,,./11335k6k88:Q;Q==?3@3BBDEGGIIKhLhNNPJQJSSU,V,XXZ[]]__aabaddfCgCiik%l%nnpqsxsxuuwZxZzz|<}<nnދލNN b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃSS SSS#:S#:S~S~SSSSpSSSSSGQS`GQSZSZSS$ SS S S SAS ASS SSlS SlSS SkSkSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations12} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations12} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations12 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 d*NSw=  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  bb)T*T44>D?DIIS0T0^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ppbb T T  FF88**  "#%%'(*y*y,,.k/k113]4]668O9O;;=A>A@@B3C3EEG%H%JJLMOOQRTxTxVVXfYf[[]T^T``b@c@eeg*h*jjlmooqqsmtmvvxSyS{{}9~9uuYYˑ˓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS˃S˃SuS uSSS:S:S~S~S%S%SSpSeSeSSS5QS`5QSqZSqZSS$ SS S S S)S )SS S;lS ;lS/S /S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_0 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1W} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_0 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1W} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_0 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1W k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Cg 1r<`+  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  XX)B*B44>+?+IIST^^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kk[[ K K  ;;++ w w""$a%a'')K*K,,.5/51134668 9 ;~;~==?h@hBBDRERGGI;J;LLN#O#QQS T VVXXZg[g]]_O`Obbd7e7ggijllnoq{q{ssucvcxxzK{K}}33yycc b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃSS SSS:S:S~S~SSSSpSSSSSQS`QS[S[SS$ SS S S S)S )SS S;lS ;lS)S )S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations13} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations13} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations13 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 xbLp6Z  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  \\)J*J44>6?6IIST^^hispsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nn\\ J J  88&& y y""$g%g'')U*U,,.C/C1131416689;;= > @@BBDrErGGI^J^LLNHOHQQS2T2VVXY[[]^`{`{bbdaeaggiGjGlln+o+qqs t v~v~xxz`{`}}BB$$vvXX b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS˃S˃SuS uSSS:S:S~S~S%S%SSpSeSeSSSQS`QSYZSYZSS$ SS S S SS SS S/lS /lS S  S߉S߉SeS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^0 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1W} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^0 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1W} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^0 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1W k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 9]\$H  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  VV)@*@44>(?(IIS T ^^hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kkWW C C  //{{e e""$O%O'')9*9,,.#/#113 4 6688:l;l==?V@VBBD@E@GGI(J(LLNOQQSSUgVgXXZM[M]]_3`3bbdeggiikmlmnnpQqQssu5v5xxz{}}ooSSƉƋ99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃS S  SSS:S:S~S~S[S[SSpSSSSSSQS`SQSZSZSS$ SM S M S;S ;SS SMlS MlSS SqSqSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 e/S_)M  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  \\)L*L44>:?:IIS&T&^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ll\\ L L  <<,,   ""$t%t'')d*d,,.T/T113D4D668494;;=$>$@@BCEEGHJzJzLLNfOfQQSRTRVVX>Y>[[]*^*``bceeghjxjxllndodqqsPtPvvx1?1IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggUU C C  11  r r""$`%`'')N*N,,.@}@}BBDkEkGGIYJYLLNGOGQQS5T5VVX#Y#[[]^``bbdvevggidjdllnMoMqqs1t1vvxy{{}}kk݂݄OO33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃS S  SSS:S:S{~S{~S[S[SSpSSSSSRS`RS=[S=[SS$ SM S M S#S #SS S5lS 5lS)S )S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 yoe[  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  VV)B*B44>,?,IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjVV B B  ..||h h""$T%T'')@*@,,.,/,11346689;z;z==?f@fBBDRERGGI%?%IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggQQ ; ;  %%nnX X""$B%B''),*,,,./11346u6u88:_;_==?I@IBBD3E3GGIJLLNOQ|Q|SSUfVfXXZP[P]]_:`:bbd$e$ggijllnnpgqgssuKvKxxz/{/}}iiۉۋMM11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYSʃSʃSS SSS:S:S{~S{~SsSsSSpSSSSS}QS`}QSZSZSS$ SM S M S)S )SS S;lS ;lS_S _S/S/SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1/} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1/} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1/ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 j>b6| Rv  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  VV)D*D44>0?0IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhVV D D  22  s s""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIXJXLLNBOBQQS,T,VVXY[[]^`u`ubbd_e_ggiIjIlln3o3qqstvvxy{|{|}}ffۂۄPPŇʼn::$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYS˃S˃SS SSS:S:S~S~SSSSpSkSkSSSAQS`AQS}ZS}ZSS$ SM S M S#S #SS S5lS 5lSYS YS)S)SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1qB b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1W} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1qB b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1W} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1qB b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1W k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ei7zHl:  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  [[)G*G44>1?1IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1gg[[ O O  CC77++  "#%%'')u*u,,.a/a113M4M668999;;=%>%@@BCEEGGIqJqLLNYOYQQSATAVVX)Y)[[]^``bbdmemggiUjUlln=o=qqs%t%vvx y {{}}ii݂݄QQŇʼn99!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYSʃSʃSS SSS:S:Su~Su~SsSsSSpSSSSSQS`QS+[S+[SS$ SM S M SS SS S#lS #lS S  S߉S߉SqS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations10} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations10} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations10 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 xj\Nr  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  SS)=*=44>%?%IIS T ^~^~hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++yye e""$P%P''):*:,,.$/$11346688:m;m==?W@WBBDAEAGGI)J)LLNOQQSSUhVhXXZN[N]]_4`4bbdeggijlslsnnpYqYssu?v?xxz%{%}} ~~dd׉׋JJ00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYS˃S˃SS SSS:S:S~S~SSSSpSkSkSSSRS`RS[[S[[SS$ SM S M SS SS S#lS #lS S  S߉S߉S_S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1pB b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1W} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1pB b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1W} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1pB b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1W k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 9]Ko] o  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  WW)?*?44>%?%IIS T ^{^{hhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggWW G G  77''  ""$k%k'')S*S,,.;/;113#4#668 9 ;;==?g@gBBDOEOGGI5J5LLNOQQSSUoVoXXZS[S]]_7`7bbdeggiikqlqnnpUqUssu9v9xxz{}}ssWWɉɋ;; b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYSʃSʃSS SSS:S:Su~Su~SOSOSSpSSSSS[RS`[RS[S[SS$ SG S G SS SS S/lS /lS5S 5SSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 g},B4X  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  XX)D*D44>,?,IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hh\\ P P  DD88((  "#%v%v'')b*b,,.N/N113:4:668&9&;;=>@@BBDlElGGITJTLLN?IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``LL 8 8  $$ppZ Z""$D%D'').*.,,./11346w6w88:a;a==?G@GBBD-E-GGIJLLNNPlQlSSURVRXXZ8[8]]_`bbdeggiikglgnnpMqMssu3v3xxz{}}rrVVȉȋ:: b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYSʃSʃSS SSS :S :So~So~SOSOSSpSSSSSqQS`qQSZSZSS$ SG S G SS SS SlS lS S  S߉S߉S;S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Jn&J&pLp  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  RR):*:44>?IIST^x^xhhrZsZ} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&&n n""$V%V'')>*>,,.&/&11346688:j;j==?N@NBBD2E2GGIJLLNNPlQlSSUPVPXXZ4[4]]_`bbdegrgriikVlVnnp:q:ssuvxxz{}t}tXXʄʆ:: b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYS˃S˃SS SSS:S:S~S~SISIS Sp SqSqSSSeQS`eQSZSZSS SG S| G SS S!~S !~SMS MS{ S { Sq݉Sq݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yK b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yK b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yK b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1weSwAe  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  KK)1*144>?IISS]n^nhhrNsN} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&nnT T""$:%:'') * ,,./1y1y335_6_88:E;E==?'@'BBD E GzGzIIK\L\NNP>Q>SSU V XXZ[]s]s__a]b]ddf?g?iik!l!nnpqststuuwVxVzz|8}8jjڋڍJJ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LSSʃSʃSS SSS:S:Si~Si~SgSgS}Sp}SSSSSQS`QS[S[SS SA S| A S/S /SW~S W~S5NS 5NSS SމSމS'S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1<p b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1,} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1<p b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1,} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1<p b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1, k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1`hp  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  VV)J*J44>:?:IIS&T&^^hiss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddVV H H  ::,,  "#%}%}'')o*o,,.a/a113S4S668E9E;;=7>7@@B)C)EEGHJJL M OOQQSnTnVVX\Y\[[]J^J``b6c6eeg h jjl m ooqqsotovvxUyU{{};~;!!ww[[͓͑ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LSS#˃S#˃SS SSS:S:S~S~SSSSpSwSwSSS%RS`%RSa[Sa[SS SA S| A S#S #SK~S K~S)NS )NS)S )SމSމSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1W} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1W} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1W k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1&JLpr)  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  LL)@*@44>)?)IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__OO ? ?  //  k k""$_%_'')I*I,,.3/311346689;|;|==?f@fBBDPEPGGI9J9LLN!O!QQS T V}V}XXZe[e]]_M`MbbdAeAggi)j)llnoqqssumvmxxzU{U}}==%%mm b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LSSʃSʃSS SSS:S:Si~Si~SmSmS}Sp}SSSSSQS`QSSZSSZSS SA S| A SS SE~S E~S#NS #NSS SމSމSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1:n b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1-} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1:n b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1-} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1:n b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1- k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1Ae+zq  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  PP)B*B44>.?.IIST^^histst} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$_%_'')M*M,,.;/;113)4)6689;;=>@|@|BBDjEjGGIVJVLLN@O@QQS*T*VVXY[[]]_s`sbbdYeYggi?j?lln/o/qqstvvxxzd{d}}FF(( zz\\ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LSS#˃S#˃SS SSS:S:S~S~SSSSpSwSwSSSQS`QSSZSSZSS SA S| A SS S9~S 9~SNS NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1W} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1W} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1W k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1mFj D  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  JJ)<*<44>$?$IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__KK 7 7  ##ooY Y""$K%K'')5*5,,./113 4 6~6~88:h;h==?R@RBBD4?4IIS"T"^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1@@..    (({{i i""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGI\J\LLNJOJQQS8T8VVX&Y&[[]^``bceyeyggigjglln\o\qqs@t@vvx$y${{}~zz^^ЇЉBB&&# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xMS5˃S5˃SSdSSlSW!SW!S$S$S:S@:S S S7S7SϐS@ϐSMS MSSSSSS SGS GSSd S CS  CS3SS 3SSo\S o\SSS῅S῅SSS S S %S  %S/S/SSpSAlStAlSySTySSSeS$eS S(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j6 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 C)~ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j6 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 C)~ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j6  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 C)~  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`(L,Ptx Dh|$H\p<t@ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@<` \@|<` \@|<` \@| < `   \   @ |  > b   & d  (LNr6CtA8\ ?^="F ;Hl 09n72V5X|@ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d<x4p,h$`XTM  (  P    L ^  9  H  D bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  SS)E*E44>5?5IIS!T!^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33## { {""$m%m'')]*],,.M/M113=4=668-9-;;==?w@wBBDgEgGGIuJuLLNaOaQQSMTMVVX9Y9[[]%^%``bceeggisjsllnkokqqsWtWvvxCyC{{}/~/}}iiߑߓ  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1MScʃScʃSS|SmSmSSS"S|"S&SH&S9S9S?~SP?~S僄SX僄ScScSSPSySySSSSSS$ S]S, ]SS SsS$ sSS S9DS| 9DSSSXSS\S|\S5S5SaʅSxaʅSS\SSSSSQ(SQ(SS8SSS#lS#lSySyS Sl S߉S߉SkS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ol b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1+ bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ol b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1+  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ol  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1+ bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`=ajNr2*g&JCv>b.R Gk b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@h 0$H<`Txl4( L  @ d  X |   p   8,PDh\ p2Cg"F9~ 6ZPtJn"F^] /Sr %4$H8\Lp` t  "!!!"6""" b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dhP8 pX@( v \ B 2 ( w  <  tZF@&  bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10??  HH):*:44>&?&IIST^^hisrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1??))    {{eeO O""$A%A'')+*+,,./11335t6t88:^;^==?H@HBBD2E2GGIJLLNOQ}Q}SSUgVgXXZQ[Q]]_;`;bbd%e%ggijllnoqtqtssuXvXxxz<{<}} vvZZ̎̐>># b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xMS5˃S5˃SSdSSlS]!S]!S$S$S:S@:SSS=S=SՐS@ՐSSS SS%S%SSSS SMS MSSd SCS CSoSS oSS\S \SSS=DžS=DžSSSESESC'S C'SSSSpS)lSt)lSySTyS#S#S߉S$߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 D* bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 D* bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 D*  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`<`}<`xLpr lLp  \ 0\ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@GkA^4X .uKo!Eb8\ 2 y   O s  % I    f  < `  2wIm?Vz(Lc5Y+pBf8} 3Os 1!E /\ -+)'% b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dGUc*q8F P   Z   d  )  n o3Jx% bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  MM)=*=44>)?)IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%ss_ _""$O%O'');*;,,.'/'11346688:u;u==?B@BBBD.E.GGI9J9LLN!O!QQS T V}V}XXZe[e]]_M`Mbbd5e5ggijllnoqqssumvmxxzU{U}}==%% ii  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSʃSʃSSdSSlSi!Si!S$S$S:S@:Si~Si~SISISᐄS@ᐄS_S _S}S}SSSS SYS YSSd SCS CSRS RS[S [S9S9ScScSSSSS$S $S#S#SSpS5lSt5lSySTySYSYS)S$)SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 8; bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 8; bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 8;  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`;_ Dt 0Ptq  /<`<?cl(Ims b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@;_X|:u3WPt2m+O Hl*e# G    @ d   " _   A |  : ^    W{Vz9,Pt&2VnOsFj 1@l*N] Gk)d"F?c b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d;v1l'bXN fD-  t  : ; w  2  m  ( c  bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  VV)B*B44>?IIST^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbVV J J  >>22&&  "#%%'')p*p,,.\/\113H4H668494;;= > @@B C EEGGIPJPLLN8O8QQS T VVX&Y&[[]^``bbdjejggiRjRllnDoDqqs,t,vvxy{{}}ppXẊ̉@@((  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSS/˃S/˃SSdSSlSi!Si!S$S$S:S@:SSSISISᐄS@ᐄS_S _SSS}S}SS SYS YSSd SCS CSSS SS\S \S S SSSSSSS"S"SS$SSS lS lSySyS/SX/S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_+ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 NU  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_+ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 NU  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_+  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 NU   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`-Q9@d.RCgHlVz,`Y} /l b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1Im#G#lFj DgAe?b< `   :  W{QuKoEi?c9]3W-Q b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r0_p64_c0_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements1800input0Elements1800invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1<I[$o8J\6  h A   bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  QQ)?*?44>-?-IIST^^hisZsZ} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccQQ ? ?  --  n n""$\%\'')J*J,,.8/8113&4&6689;;=>@y@yBBDgEgGGIUJULLN?O?QQS)T)VVXY[[]]_r`rbbd\e\ggiFjFlln2o2qqstvvxy{{{{}}eeڂڄOOzz  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSʃSʃSSdSSlSi!Si!S$S$S:S@:Si~Si~SISISᐄS@ᐄS_S _S}S}SSSS SYS YSSd SCS CSSS SS\S \SGSGS9…S9…SSSSS%S %S S SSpSlStlSySTyS S S߉S$߉SYS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 7; bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 7; bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1;o  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 7;  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`@dEi;Pt  .Rvn(L-Q~8\Vz b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@;_X|:v4XQu3n,PIm+f$ H    A e   # ^   @ {  9 ]   Vz8s1UNr 0k)M Fj3W( -ct!EJn D>ba7[ 1x b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d;v1l'bYO E  = x 3 I  n  )  d W  g bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  RR):*:44>?IIST^^hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbRR B B  22"" z z""$f%f'')N*N,,.6/611346689;z;z==?b@bBBDJEJGGIJLLNNPiQiSSUlVlXXZP[P]]_4`4bbdeggiiknlnnnpZqZssu>v>xxz"{"}}xx\\Ή΋@@$$  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSS/˃S/˃SSdSSlSi!Si!S$S$S:S@:S S SISISᐄS@ᐄS_S _SSSSSS SYS YSSd SCS CSSS SS\S \S S SƅSƅSi Si SSS'S'S穆S$穆SSSkSkSsySsyS SX S߉S߉SAS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a- b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 OU  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a- b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 OU  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a-  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 OU   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`=aoMq@d+OSw) -Vzx,i?l b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1;_X|:u3WPt2n,PIm+f$ H 'K!E?93 -'! b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r0_p64_c0_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements1800input0Elements1800invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1<;v1l'bXN s L%~W bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  NN)8*844>"?"IIST^^hhrFsF} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&tt` `""$K%K'')5*5,,./113 4 6~6~88:h;h==?R@RBBD?IIST^u^uhhr=s=} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__KK 7 7  ##ooY Y""$C%C'')-*-,,./11346v6v88:`;`==?H@HBBD.E.GGIJLLNNPmQmSSUSVSXXZ9[9]]_`bbdegxgxiik^l^nnpDqDssu*v*xxz{}}ii܄܆11  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lSS#˃S#˃SSLSSTS{!S{!S$S$S:S(:S~S~S[S[SS(SqSqSSSwSwSS SkS kSSL S1CS 1CSSS SS\S \SSSSSSSSDSSS)lS )lS5S5SS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Y  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Y  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Y   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Fj.RBDh^Jn"2V`8f>| b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@,"yoe[QuGk bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 j A  o  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  WW)C*C44>-?-IIST^^hhrqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1gg[[ O O  CC77''  ""$u%u'')a*a,,.M/M113949668%9%;;=>@@BBDmEmGGIUJULLN=O=QQS%T%VVX Y [[]]_i`ibbdQeQggi9j9lln!o!qqs t v}v}xxze{e}}MM55uu b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSʃSʃSSdSSlSu!Su!S$S$S :S@ :So~So~SUSUS퐄S@퐄SkS kSSSSSS SeS eSSd S+CS +CSSS SS]S ]S_S_S…S…S[S[SSS%S%SϩS$ϩSSSkSkS[yS[ySQ SXQ S!߉S!߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1<p b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ; bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1<p b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ; bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1<p  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ;  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Cg#EiUy7z:^6ZGk0T'j b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1Rv>b*N:&xdPt<`(L 8   $ v   b b<`:]7[5~ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r0_p64_c0_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements1800input0Elements1800invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1<R$vHl>b4 ^ ' p 9 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  JJ)0*044>?IISS]g^ghhr*s*} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%mmS S""$9%9'')*,,./1x1x335^6^88:D;D==?(@(BBD E G{G{IIK]L]NNP?Q?SSU!V!XXZ[]t]t__aVbVddf8g8iiklnnpprmsmuuwOxOzz|1}1ֆֈFF&&  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lSS)˃S)˃SSLSSTS{!S{!S$S$S:S(:S~S~S[S[SS(SqSqSSSwSwSS SkS kSSL S1CS 1CSSS SSW\S W\SSS;ƅS;ƅSSSSDSSSlS lS S S߉S0߉S5S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V" b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Z  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V" b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Z  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V"  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Z   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`i6Z$H~4"Fl*N4j2DhX|  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@=a3W)MC9 /%|r bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 F tK" } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  QQ)9*944>?IIST^u^uhhrYsY} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%%m m""$U%U'')=*=,,.%/%113 4 6688:i;i==?O@OBBD3E3GGIJLLNNPmQmSSUQVQXXZ5[5]]_`bbddfogoiikSlSnnp7q7ssuvxxzz|q}qUUDŽdž99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLSSTS!S!S%S%S:S(:Su~Su~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CSRS RS[S [SSS9S9SSSSDSSS)lS )lS#S#S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 -=  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 -=  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 -=   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`@al!#G=a AFj],=a} 1] b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ ];_=fDh"F$o bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1|G  ] (  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  OO)9*944>$?$IIS T ^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbRR B B  22""n n""$X%X'')B*B,,.,/,11346689;u;u==?_@_BBDIEIGGI4J4LLNOQQSTVxVxXXZ`[`]]_H`Hbbd0e0ggijllnoqtqtssu\v\xxzD{D}},,tt^^ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS˃S˃SSLSSTS!S!S%S%S:S(:S~S~SsSsS S( SSS S SqSqSS SS SSL SICS ICS-SS -SSi\S i\SSSSSSS/SD/SSSAlS AlSSSYS0YSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zI b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 dU. bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zI b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 dU. bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zI  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 dU.  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`;_'K+O(Ls==t+>e~ /f 0 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@d2V$H:,seW{ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1rG  q F  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  YY)K*K44>=?=IIS)T)^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggYY K K  ==//!!  "#%~%~'')p*p,,.b/b113T4T668F9F;;=8>8@@B*C*EEGHJJL M OOQQSqTqVVX_Y_[[]M^M``b9c9eeg#h#jjl m ooqqsftfvvxLyL{{}2~2ppTTƑƓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLSSTS!S!S%S%S :S( :So~So~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CSSS SS\S \S#S#S)ÅS)ÅSSS SD SSSlS lS S S߉S0߉SSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ,< bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ,< bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ,<  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`?c{!Ei%[)3a -9w+O} b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Mq+O -x Vz4X6_ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 s > T } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  MM)7*744>!?!IIST^x^xhhr^s^} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&rr\ \""$F%F'')0*0,,./11346y6y88:c;c==?M@MBBD7E7GGI!J!LLNOQzQzSSU`V`XXZF[F]]_,`,bbdeggiikflfnnpJqJssu.v.xxz{}}hhڄچNN44 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS#˃S#˃SSLSSTS!S!S%S%S:S(:S~S~SySySS(SSS S SqSqSS SS SSL SOCS OCSmRSZmRS[S[SS SŅS ŅSSSSSSS)lS)lSSS߉S*߉SSJ} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{K b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 eU0 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{K b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 eU0 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{K  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 eU0  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z1z5*5z9*9b=reasonOperatorKernelvalue7kernelSpecoutputElements1152input0Elements1152invocations11bPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Fj+Oj>Gk8Z~c.R0T 1Lp b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Im;_-QC5'|n  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1  p Eo } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>/?/IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIWJWLLNAOAQQS+T+VVXY[[]]_t`tbbdZeZggi@j@lln$o$qqstvwvwxxzY{Y}};;qqSS b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTS!S!S)%S)%S:S(:S{~S{~SSS#S(#SSSSSSSS SS SSL SaCS aCSQS QS1[S 1[SySySSS S S#SD#SSS5lS 5lSMSMSS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  S@ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  S@ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  S@  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`7Z~v5fX| 1TxTx -BfPt b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@'K -z\>b D&sUy bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1T ! n ;  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  PP)>*>44>.?.IIST^^hisrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$[%['')I*I,,.7/7113%4%6689;;=>@x@xBBDfEfGGIVJVLLNDODQQS2T2VVX Y [[]^``bbdsesggiajallnJoJqqs.t.vvxy{{}}hhڂڄLL00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S˃S˃S{SL{SSTS!S!S %S %S:S(:S~S~SmSmSS(SSSSSkSkSS S}S }SSL SCCS CCSRS RS[S [SESESSSqSqS5SD5S S SGlS GlSSSkS0kSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o>  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 d~" bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o>  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 d~" bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o>   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 d~"  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`;_ Dt5 0Pt|  <`Tx*N<l(Hl b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@jX|Fj4X"F4"u bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1Q$ w J   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  WW)G*G44>7?7IIS#T#^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggWW G G  77''  ""$o%o'')_*_,,.O/O113?4?668/9/;;=>@@BCEEGGIwJwLLNcOcQQSOTOVVX;Y;[[]'^'``bceeggiujullnaoaqqsMtMvvx9y9{{}%~%ss__ՑՓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTS!S!S)%S)%S:S(:Su~Su~SSS#S(#SSSSSSSS SS SSL SaCS aCSRS RS\S \SWSWSÅSÅS S SSDSSS#lS #lS S S߉S0߉SkS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  R> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  R> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  R>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`2Vcw$H5f $ImEi 04Xy7h b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@7[=lNr0T6e bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 U "o< } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  LL)6*644>"?"IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbLL 6 6      iiS S""$=%='')'*',,./11335p6p88:Z;Z==?D@DBBD.E.GGIJLLNOQyQySSUcVcXXZM[M]]_7`7bbd!e!ggi j llnnpdqdssuHvHxxz,{,}}ff؉؋JJ.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S˃S˃S{SL{SSTS!S!S %S %S:S(:S~S~SmSmSS(SSSSSkSkSS S}S }SSL SCCS CCSQS QSZS ZSCSCS-ŅS-ŅSSSSDSSS/lS /lS#S#S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p@  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 e$ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p@  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 e$ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p@   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 e$  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`;_ DTx4X|Dh:%5Ywy%Ii b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@cQu?c-Q? -n \ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1p C i< } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  QQ)=*=44>)?)IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))wwc c""$O%O'');*;,,.'/'11346688:u;u==?a@aBBDMEMGGI9J9LLN!O!QQS T V}V}XXZe[e]]_M`Mbbd5e5ggijllnoqyqyssuavaxxzI{I}}11uu]] b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃS SL SSTS{!S{!S$S$S:S(:S{~S{~S[S[SS(SqSqSSSSSS SkS kSSL S1CS 1CSQS QS [S  [SUSUSkSkSMSMS)SD)SSS;lS ;lS_S_S/S0/SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ";  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ";  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ";   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` .HlovCg@dn9 6Z~d ,PPtZ~ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@;)|jX|Fj4X"F bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1*} P # v  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  ZZ)F*F44>2?2IIST^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffZZ N N  BB66**  "#%%'')t*t,,.`/`113L4L668898;;=$>$@@BCEEGGIrJrLLNZOZQQSBTBVVX*Y*[[]^``bbdnenggiVjVlln>o>qqs&t&vvxy{{}}jjނބRRƇƉ::"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S˃S˃SuSLuSSTSu!Su!S$S$S:S(:S~S~SUSUS퐄S(퐄SkSkSSSeSeSS SeS eSSL S+CS +CS7RS 7RSs[S s[SSSSSSSMSMS/SD/SSSAlS AlSeSeS5S05SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d3 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 MY bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d3 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 MY bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d3  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 MY  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`;_=b $%Ian#G3W| &?c2 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@?)zdNr8\"F 0 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations10 R # t  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  UU)C*C44>1?1IIST^^hiszsz} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggUU C C  11  r r""$`%`'')N*N,,.@}@}BBDkEkGGIYJYLLNCOCQQS-T-VVXY[[]^`v`vbbd`e`ggiJjJlln4o4qqstvvxy{}{}}}gg܂܄QQƇƉ;;%% b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTS{!S{!S$S$S:S(:S{~S{~S[S[SS(SqSqSSSSSS SkS kSSL S1CS 1CS RS  RSI[S I[SSSąSąSmSmSSDSSS)lS )lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 !;  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 !;  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 !;   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$Hf\(L&J^(LKo` 1*Nb b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@4"ucQu?c-Q? - bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1I  oB } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  VV)>*>44>&?&IIS T ^|^|hhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&& ~ ~""$j%j'')R*R,,.:/:113"4"668 9 ;~;~==?f@fBBDNENGGI6J6LLNOQQSSUpVpXXZT[T]]_8`8bbdeggijlrlrnnpVqVssu:v:xxz{}}ttXXʉʋ<<  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S˃S˃SuSLuSSTSu!Su!S$S$S:S(:S~S~SUSUS퐄S(퐄SkSkSSSeSeSS SeS eSSL S+CS +CSQS QSZS ZSSSąSąSaSaSSDSSS)lS )lS S S߉S0߉SwS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e5 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 LZ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e5 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 LZ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e5  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 LZ  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`;_9>b#X|(Lt :6u`"Jns b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@kUy?c)M7!r \ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1E  g8  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  RR)<*<44>&?&IIS T ^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **xxd d""$O%O'')9*9,,.#/#113 4 6688:l;l==?V@VBBD@E@GGI*J*LLNOQQSSUiViXXZO[O]]_5`5bbdeggijltltnnpZqZssu@v@xxz&{&}} ee؉؋KK11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSSLSSTS!S!S$S$S:S(:S~S~SaSaSS(SwSwSSSSSS SqS qSSL S7CS 7CSQS QS [S  [SUSUSuSuSSSSDSSS)lS )lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 <  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 <  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 <   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`%=a6Zn<0TUy*NfQuDh(LMq b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@%) - 159=A bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 v T 2   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  NN)8*844> ? IIST^y^yhhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''ss] ]""$G%G'')1*1,,./11346z6z88:d;d==?L@LBBD2E2GGIJLLNNPqQqSSUWVWXXZ=[=]]_#`#bbd e g|g|iikblbnnpHqHssu.v.xxz{}}mmSSʼnŋ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS ˃S ˃SoSLoSSTS!S!S$S$S}:S(}:S~S~SaSaSS(SwSwSSSYSYSS SqS qSSL S7CS 7CSRS RSO[S O[SSSSSSS/SD/SSSAlS AlSGSGSS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y( b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 P u bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y( b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 P u bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y(  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 P u  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`6Pt| -`Vz3Dh\(L9b  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@th\PtDh9]-Q bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1pF  r H  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  [[)G*G44>1?1IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kk__ S S  GG;;++  "#%y%y'')e*e,,.Q/Q113=4=668)9);;=>@@BBDqEqGGIYJYLLNAOAQQS)T)VVXY[[]]_m`mbbdUeUggi=j=lln%o%qqs t vvxxzi{i}}QQłń99 yy b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS SL SSTS!S!S$S$S:S(:S~S~SaSaSS(SwSwSSSSSS SqS qSSL S7CS 7CSQS QS%[S %[SmSmSDžSDžS?S?SSDSSSlS lS S SW߉S0W߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 < bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 < bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 <  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`<DhEi4:vAe8Qu?c/S8?c b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@!E%I)M-Q1U5Y9]=a bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 n L*f } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  NN)4*444>?IISS]k^khhrMsM} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))qqW W""$=%='')#*#,,. / 1|1|335b6b88:H;H==?,@,BBDEGGIIKaLaNNPCQCSSU%V%XXZ[]x]x__aZbZddf#?#IIST^y^yhhr]s]} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiYY I I  99))q q""$Y%Y'')A*A,,.)/)11346688:m;m==?S@SBBD7E7GGIJLLNNPqQqSSUUVUXXZ9[9]]_`bbdegsgsiikWlWnnp;q;ssuvxxz{}u}uYY˄ˆ== b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃSSLSSTS!S!S%S%S#:S(#:S~S~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CSQS QS[S [S[S[SSSSS)SD)SSS;lS ;lS5S5SS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ; bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ; bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ;  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`;_&.RyC,Pg'*NuUy Y}(LLp b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@`z+!E;_Uyo : bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 o X A *  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  SS)=*=44>(?(IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&&r r""$\%\'')F*F,,.0/011346689;y;y==?c@cBBDMEMGGI8J8LLN O QQSTV|V|XXZd[d]]_L`Lbbd4e4ggijllnoqxqxssu`v`xxzH{H}}00xxbb b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS˃S˃SiSLiSSTS!S!S%S%Sw:S(w:S~S~SgSgSS(S}S}S梅S梅SSSSSS SwS wSSL S=CS =CSQS QS1[S 1[SySySSSCSCSASDASSSSlS SlSSSkS0kSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Sd bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Sd bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Sd  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` />bu%Sw@d|3WBfr4Dhs b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@eoy",6@&J bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1^ ?  b  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  ]])O*O44>A?AIIS-T-^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kk]] O O  AA33%%  " # %%'')t*t,,.f/f113X4X668J9J;;=<><@@B.C.EEG H JJLMOOQQSuTuVVXcYc[[]Q^Q``b=c=eeg'h'jjlmooqqsjtjvvxPyP{{}6~6ttXXʑʓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃSSLSSTS!S!S%S%S#:S(#:S~S~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CSQS QS[S [SOSOSOʅSOʅSSSSDSSS/lS /lS S S߉S0߉SkS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ; bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ; bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ;  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`;_ ..R#G((L /##G}Bgw b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@0TJnd~ /%I? c Y!}!! "s"" bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 |eN7 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  QQ);*;44>%?%IIS T ^|^|hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **vv` `""$J%J'')4*4,,./11346}6}88:g;g==?Q@QBBD;E;GGI%J%LLN O Q~Q~SSUdVdXXZJ[J]]_0`0bbdeggiikjljnnpNqNssu2v2xxz{}}llބކRRʼnŋ88 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS˃S˃SiSLiSSTS!S!S%S%S}:S(}:S~S~SgSgSS(S}S}SSSYSYSS SwS wSSL S=CS =CSQS QS [S  [SUSUSȅSȅSSS)SD)SSS;lS ;lS)S)S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 R e bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 R e bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 R e  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H|r,`VzzDh:^^(LBBf b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@0T:^DhNrX|b l v ! bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1C $fG } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  WW)E*E44>3?3IIST^^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiWW E E  33!!t t""$b%b'')P*P,,.>/>113,4,6689;;=>@@BBDmEmGGI[J[LLNEOEQQS/T/VVXY[[]^`x`xbbd^e^ggiDjDlln(o(qqs t v{v{xxz]{]}}??!!uuWW b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS ˃S ˃SoSoSS$S}:S}:S~S~S[S$[SSSYSYSS$SkQSkQSZS4ZSST S_ S _ SGS GSS SYlS YlSwS wSGSGSS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19mS! b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19mS! b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19mS! b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 >b'rPt.R  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ww  ee)S*S44>A?AIIS/T/^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1wwee S S  AA//   ""$p%p'')^*^,,.L/L113:4:668(9(;;=>@@BCE{E{GGIiJiLLNWOWQQSETEVVX3Y3[[]!^!``bceeggitjtlln]o]qqsAtAvvx%y%{{} ~ {{__чщCC'' b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃScS cSSSw:Sw:S~S~S[S[S梅Sp梅SSSSSSSQS`QSZSZSS$ S_ S _ SSS SS+S +SelS elSS SSS1S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 oMq+O -x  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  dd)T*T44>B?BIIS.T.^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ttdd T T  DD44$$  "#%|%|'')l*l,,.\/\113L4L668<9<;;=,>,@@BCEEG H JJLLNnOnQQSZTZVVXFYF[[]2^2``bceeg h jjllnlolqqsXtXvvxDyD{{}0~0~~jj b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS ˃S ˃SoSoSS$S:S:S~S~S[S$[SSS_S_SS$SqQSqQSZS4ZSST S_ S _ S5S 5S S  SGlS GlSS S߉S߉SS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1:nQ$ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1:nQ$ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1:nQ$ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ;_ .y W{5Y  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10uu  __)I*I44>3?3IIST^^hissss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1uu__ I I  33||f f""$P%P''):*:,,.$/$11346688:m;m==?W@WBBDAEAGGI+J+LLNOQQSSUtVtXXZ^[^]]_H`Hbbd2e2ggijllnoququssuYvYxxz={=}}!!ww[[͎͐?? b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃScS cSSSw:Sw:S~S~S[S[S梅Sp梅SSSSSSSQS`QSZSZSS$ S_ S _ SAS ASS SSlS SlSGS GSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 tRv0T2}  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  ``)L*L44>6?6IIST^^hiszsz} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1tt`` L L  88$$r r""$^%^'')J*J,,.6/6113"4"6689;;==?p@pBBD\E\GGIFJFLLN.O.QQSTVVXXZr[r]]_Z`ZbbdBeBggi*j*llnoqqssunvnxxzV{V}}>>&&jj b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS'S 'SSS5:S5:S~S~SmSmSSpSSSSSeQS`eQSZSZSS$ S_ S _ SGS GSS SYlS YlS}S }SMSMSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ei;Sw-Q  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ww  kk)W*W44>A?AIIS)T)^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1wwkk _ _  SSGG;;//  "###%%'(**,,.q/q113]4]668I9I;;=5>5@@B!C!EEG H JJLLNiOiQQSQTQVVX9Y9[[]!^!``b c e}e}ggiejellnMoMqqs5t5vvxy{{}~yyaaՇՉII11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS˃S˃SiS iSSSw:Sw:S~S~S=S=S梅Sp梅SSSSSSSQS`QSZSZSS$ S_ S _ SAS ASS SSlS SlSwS wSGSGSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1d} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1d} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1d k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 oMq+O -x  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  ^^)L*L44>8?8IIS"T"^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp^^ L L  ::(( { {""$i%i'')W*W,,.E/E113343668!9!;;=>@@BBDtEtGGI`J`LLNJOJQQS4T4VVXY[[]^`}`}bbdgegggiQjQlln;o;qqs%t%vvxy{{}}nnXX͇͉BB,, b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS!S !SSS5:S5:S~S~SmSmSSpSSSSSqQS`qQSZSZSS$ S_ S _ S/S /SS SAlS AlSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ;_ 0{Y}7[  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10uu  ee)M*M44>3?3IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1uuee U U  EE55%%  "#%y%y'')a*a,,.I/I1131416689;;=>@u@uBBD]E]GGICJCLLN'O'QQS T V}V}XXZa[a]]_E`Ebbd)e)ggi j llnnpcqcssuGvGxxz+{+}}ee׉׋II-- b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS˃S˃SiS iSSSw:Sw:S~S~S=S=S梅Sp梅SSSSSSSQS`QSZSZSS$ S_ S _ S/S /SS SAlS AlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1e} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1e} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1e k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 uSw1U3~  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  ]])G*G44>/?/IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qq]] I I  55!!  o o""$Z%Z'')D*D,,../.11346689;w;w==?a@aBBDKEKGGI3J3LLNOQQSSUrVrXXZX[X]]_>`>bbd$e$ggi j l}l}nnpcqcssuIvIxxz/{/}}nnTTǎǐ:: b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSS SSS/:S/:S~S~S[S[SSpS S SSSYQS`YQSZSZSS$ SY S Y S#S #SS S5lS 5lS S  S߉S߉SqS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 f.RZ~"F  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  [[)E*E44>+?+IIST^^hhrhsh} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp\\ H H  44    j j""$T%T'')>*>,,.(/(11346688:q;q==?W@WBBD=E=GGI#J#LLN O Q|Q|SSUbVbXXZH[H]]_.`.bbdeggiikmlmnnpSqSssu9v9xxz{}}xx\\Ή΋@@$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS ˃S ˃SoS oSSS}:S}:S~S~S[S[SSpSYSYSSS#QS`#QS_ZS_ZSS$ SY S Y S5S 5S S  SGlS GlSMS MSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T# b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T# b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T# b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ?c%b$H  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  ``)L*L44>4?4IIST^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ppdd X X  LL@@00  "#%~%~'')j*j,,.V/V113B4B668.9.;;=>@@BCEtEtGGI\J\LLNDODQQS,T,VVXY[[]]_p`pbbdXeXggi@j@lln(o(qqstvvxxzl{l}}TTȂȄ<<!!zz b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSS SSS):S):S~S~S[S[SSpSSSSSGQS`GQSZSZSS$ SY S Y SS SS S#lS #lS S  Sc߉Sc߉S S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 h*NMq3  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  YY)?*?44>!?!IIST^t^thhrTsT} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp\\ H H  44    ||b b""$H%H'').*.,,./11335m6m88:S;S==?5@5BBDEGGIIKjLjNNPLQLSSU.V.XXZ[]]__acbcddfEgEiik'l'nnp q szszuuw\x\zz|>}> ppPP b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS ˃S ˃SoS oSSS}:S}:S~S~S[S[SSpSYSYSSS QS` QSGZSGZSS$ SY S Y SS SS S/lS /lS S  S߉S߉SMS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S# b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S# b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S# b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ;_Z~<w  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  \\)D*D44>(?(IIS T ^~^~hhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp`` P P  @@00 x x""$`%`'')H*H,,.0/011346689;t;t==?X@XBBD[>]]_"`"bbdegxgxiik\l\nnp@q@ssu$v$xxz{}z}z^^ЄІ@@"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃSS SSS):S):S~S~SSSSpSSSSSAQS`AQS}ZS}ZSS$ SS S S S)S )SS S;lS ;lS5S 5SSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1.} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1.} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1. k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 f(LKo 1  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10oo  \\)F*F44>/?/IIST^^hhrsss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1oo__ O O  ??// { {""$e%e'')O*O,,.9/9113#4#668 9 ;;==?l@lBBDVEVGGI?J?LLN'O'QQSTVVXXZk[k]]_S`Sbbd;e;ggi#j#lln o qqssugvgxxzO{O}}77}}gg b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS ˃S ˃SoS oSSS:S:S~S~S%S%SSpS_S_SSSQS`QSAZSAZSS$ SS S S S;S ;SS SMlS MlSS SeSeS S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1], b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1U} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1], b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1U} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1], b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1U k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ;_X|:u  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  ^^)P*P44>@?@IIS,T,^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ll^^ P P  BB44&&  " # %%'')u*u,,.g/g113Y4Y668K9K;;==>=@@B/C/EEG!H!JJLMOOQQStTtVVXbYb[[]P^P``b*?*IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mmYY E E  11}}g g""$Q%Q'');*;,,.%/%11346688:n;n==?X@XBBDBEBGGI*J*LLNOQQSSUiViXXZO[O]]_5`5bbdeggiikolonnpSqSssu7v7xxz{}}qqUUȉȋ;;!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS ˃S ˃SoS oSSS:S:S~S~S%S%SSpS_S_SSSwQS`wQSZSZSS$ SS S S S)S )SS S;lS ;lS#S #S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\, b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1U} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\, b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1U} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\, b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1U k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ;_>gEi  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  ZZ)H*H44>4?4IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1llZZ H H  66$$ w w""$e%e'')S*S,,.A/A113/4/6689;;= > @@BBDpEpGGI\J\LLNFOFQQS0T0VVXY[[]^`y`ybbd_e_ggiEjElln)o)qqs t v|v|xxz^{^}}@@""ttVV b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃSS SSS#:S#:S~S~S[S[SSpSSSSS;QS`;QSwZSwZSS$ SS S S S/S /SS SAlS AlSYS YS)S)SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 b&JLp4  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  YY)G*G44>5?5IIS#T#^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kkYY G G  55##v v""$d%d'')R*R,,.@/@113.4.6689;;= > @@BBDoEoGGI]J]LLNKOKQQS9T9VVX'Y'[[]^``bcezezggihjhllnQoQqqs5t5vvxy{{}}ooSSŇʼn77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S˃S˃SuS uSSS:S:S~S~SOSOSSpSeSeSSS#QS`#QS_ZS_ZSS$ SS S S S;S ;SS SMlS MlSS SqSqSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f5 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f5 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f5 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ?c)h.R  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  XX)H*H44>6?6IIS"T"^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhXX H H  88((  ""$p%p'')`*`,,.P/P113@4@668090;;= > @@BCEEGHJvJvLLNbObQQSNTNVVX:Y:[[]&^&``bceeggitjtlln`o`qqsLtLvvx8y8{{}$~$rr^^ԑԓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃS S  SSS:S:S~S~S[S[SSpSSSSSRS`RSO[SO[SS$ SS S S SS SS S/lS /lS S  S߉S߉SwS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 |wqk   b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  SS)=*=44>'?'IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiSS = =  ''ppZ Z""$D%D'').*.,,./11346w6w88:a;a==?K@KBBD5E5GGIJLLN O Q~Q~SSUhVhXXZR[R]]_<`*?*IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,zzf f""$R%R'')>*>,,.*/*11346689;x;x==?d@dBBDPEPGGI:J:LLN"O"QQS T V~V~XXZf[f]]_N`Nbbd6e6ggijllnoqzqzssubvbxxzJ{J}}22vv^^ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃS S  SSS:S:S~S~SsSsSSpSSSSSeQS`eQSZSZSS$ SM S M S)S )SS S;lS ;lS_S _S/S/SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1+} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1+} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1+ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 i7[)l:^  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  __)K*K44>5?5IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kk__ S S  GG;;//##  "#%%'(*y*y,,.e/e113Q4Q668=9=;;=)>)@@BCEEGHJuJuLLN]O]QQSETEVVX-Y-[[]^``bbdqeqggiYjYllnAoAqqs)t)vvxy{{}}mmUUɇɉ==%% b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S˃S˃S{S {SSS:S:S~S~SSSSpSkSkSSSYQS`YQSZSZSS$ SM S M S#S #SS S5lS 5lS_S _S/S/SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o>  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1U} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o>  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1U} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o>  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1U k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Hl@Z~.R  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  RR)@*@44>,?,IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddRR @ @  ..  o o""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGITJTLLN>O>QQS(T(VVXY[[]]_q`qbbd[e[ggiEjElln/o/qqstvvxy{x{x}}bbׂׄLL66  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃSS SSS:S:S{~S{~SsSsSSpSSSSSRS`RS[[S[[SS$ SM S M SS SS S#lS #lS S  S߉S߉S_S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1,} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1,} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1, k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 {ywu  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  YY)A*A44>'?'IIS T ^}^}hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiYY I I  99))   ""$m%m'')U*U,,.=/=113%4%668 9 ;;==?i@iBBDQEQGGI7J7LLNOQQSSUqVqXXZU[U]]_9`9bbdeggijlslsnnpWqWssu;v;xxz{}}uuYYˉˋ==!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S˃S˃S{S {SSS:S:S~S~SSSSpSkSkSSSRS`RSU[SU[SS$ SM S M SS SS S#lS #lS S  S߉S߉SkS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n>  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1U} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n>  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1U} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n>  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1U k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ;_ImW{g  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  QQ);*;44>#?#IIS T ^|^|hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))wwc c""$N%N'')8*8,,."/"113 4 6688:k;k==?U@UBBD?E?GGI'J'LLN O QQSSUfVfXXZL[L]]_2`2bbdeggiikqlqnnpWqWssu=v=xxz#{#}} ||bbՉՋHH.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYSʃSʃSS SSS:S:Su~Su~SOSOSSpSSSSSRS`RS[S[SS$ SG S G S S  SS SlS lS S  S߉S߉SSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 r*.RVz~  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  OO)9*944>?IIST^z^zhhr^s^} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((tt^ ^""$H%H'')2*2,,./11346{6{88:e;e==?K@KBBD1E1GGIJLLNNPpQpSSUVVVXXZ<[<]]_"`"bbd e g}g}iikclcnnpIqIssu/v/xxz{}}nnRRĉċ66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYS˃S˃SS SSS:S:S~S~SCSCS Sp SqSqSSSRS`RS[[S[[SS SG S| G S#S #SK~S K~S)NS )NS;S ;S1މS1މSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xG b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xG b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xG b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1"FDhf  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  TT)@*@44>(?(IIST^^histst} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddXX L L  @@44$$  ""$r%r'')^*^,,.J/J113646668"9";;= > @@BBDhEhGGIPJPLLN8O8QQS T VVXY[|[|]]_d`dbbdVeVggi>j>lln&o&qqstvvxxzj{j}}RRƂƄ::xx b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYSʃSʃSS SSS:S:Su~Su~SOSOSSpSSSSSQS`QSZSZSS$ SG S G SS SS S lS  lS{ S { SK߉SK߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 SwAe/SA  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  MM)3*344>?IISS]l^lhhrLsL} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((ppV V""$<%<'')"*",,./1{1{335a6a88:G;G==?)@)BBD E G|G|IIK^L^NNP@Q@SSU"V"XXZ[]u]u__a[b[ddf=g=iiklnnpqsrsruuwTxTzz|6}6hh؋؍HH b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYS˃S˃SS SSS:S:S~S~SCSCS Sp SqSqSSS/QS`/QSkZSkZSS SG S| G S S  S3~S 3~SNS NS S  S݉S݉SaS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wG b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wG b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wG b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1pNr,P /  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP)8*844>?IIST^z^zhhr\s\} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$l l""$T%T'')<*<,,.$/$113 4 6688:h;h==?L@LBBD0E0GGIJLLNNPjQjSSUNVNXXZ2[2]]_`bbdegtgtiikXlXnnpr b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1(} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1(} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1( k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1i4*N  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  PP)<*<44>%?%IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33##o o""$[%['')E*E,,.///11346689;x;x==?b@bBBDLELGGI5J5LLNOQQSTVyVyXXZa[a]]_I`Ibbd=e=ggi%j%lln o qqssuivixxzQ{Q}}99!! ii b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LSS#˃S#˃SS SSS:S:S~S~SSS Sp SqSqSSSQS`QS%[S%[SS SG S| G S5S 5S]~S ]~S;NS ;NSS SމSމS-S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1U} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1U} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1U k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations18(L<`  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  RR)N*N44>>?>IIS*T*^^hiss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``RR D D  66((   ""$%'')s*s,,.e/e113W4W668I9I;;=;>;@@B-C-EEGHJJL M OOQQSrTrVVX`Y`[[]N^N``b:c:eeg$h$jjlmooqrtstsvvxYyY{{}?~?%% {{__ёѓ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LSSʃSʃSS SSS :S :So~So~SgSgSSpSSSSS)QS`)QSeZSeZSS SG S| G SS S9~S 9~SNS NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1<p b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1)} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1<p b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1)} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1<p b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1) k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1Jn%I)M+  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  LL):*:44>"?"IIST^^hhrisi} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%qq[ [""$I%I'')3*3,,./11346|6|88:f;f==?P@PBBD:E:GGI"J"LLNOQ{Q{SSUaVaXXZG[G]]_-`-bbdeggijlslsnnpWqWssu;v;xxz{}}uuYỶ̋??%% b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LSS#˃S#˃SS SSS:S:S~S~SSS Sp SqSqSSSPS`PS)ZS)ZSS SG S| G SS SE~S E~S#NS #NSS SމSމSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1U} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1U} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1U k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1oAe+j  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  NN)D*D44>0?0IIST^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``NN < <  **}}k k""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIXJXLLNBOBQQS,T,VVXY[[]^`u`ubbd[e[ggiAjAlln1o1qqstvvxxzf{f}}HH** ||^^ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations1MSiʃSiʃSS|SsSsSSS"S|"S&SH&S9S9SE~SPE~S냄SX냄SiSiSSPSSSYSYSS$ ScS, cSS SyS$ ySS S?DS| ?DSQSXQS[S|[SISISaSxaSS\SSS%S%S5S85S S SGlSGlSySySSlS_S_SS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Pq b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1- bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Pq b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1-  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Pq  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1- bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`!EBfVz3:wKo  .Y}?c"<`3Wx b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@5YBf2'Kt=@d 0 /r!>bv .ph<`Z~,Lpn:^ * l  8 \   ( j  6 Z   ( l  <` 0tDh8|Lp@Tx$H\,P d4X b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d5B F_4J  NRX ` $  h  , p 4  x< bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10@@  RR)N*N44>@?@IIS,T,^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1@@22 $ $  66((   ""$%'')s*s,,.e/e113W4W668I9I;;=;>;@@B-C-EEGHJJLMOOQQStTtVVXbYb[[]P^P``b b   $ a  # G   Fj,i+ONr4q3WVz<y;_!^ D b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d=?|9svN3)p-j'd"_  Y   S   M  G bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  YY)G*G44>3?3IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeYY M M  AA55))  "#%%'')u*u,,.a/a113M4M668999;;=>@{@{BBDgEgGGIsJsLLN[O[QQSCTCVVX+Y+[[]^``bbdoeoggiWjWllnKoKqqs3t3vvxy{{}~ww__ӇӉGG//  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1MSiʃSiʃSS|SsSsSSS"S|"S&SH&S9S9S?~SP?~S냄SX냄SiSiSSPSSSSSSSS$ ScS, cSS SyS$ ySS S?DS| ?DSRSXRS[S|[S3S3SͅSxͅSS\S1S1S/)S/)SS8SSS/lS/lSySySSlS߉S߉SS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1On b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1- bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1On b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1-  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1On  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1- bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Bsp .(Fj8\,Pn&fTx! *N} b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@s .8\f!+OY}B>bL p 0 T   " F z  5  8  ? c  *    m q ( c2V`%IQu {4:^d#GMqw 06Z` C  I!m!!"s""#,###2$V$$$\%% b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dsfYL?^23%]2 ~  o  `  Q  B  3$wh bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10??  NN)D*D44>2?2IIST^^hiszsz} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1??--    **}}k k""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIZJZLLNDODQQS.T.VVXY[[]^`w`wbbd]e]ggiCjClln3o3qqstvvxxzh{h}}JJ,,bb# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xMS5˃S5˃SSdSSlS]!S]!S$S$S:S@:S S S=S=SՐS@ՐSSS SSSSSSS SMS MSSd SCS CSQS QSZS ZS=S=S῅S῅SSS S S %S  %SSSSpS#lSt#lSySTyS S S߉S$߉S_S(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h4 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 B& bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h4 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 B& bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h4  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 B&  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`?rUy 1X|Gk#=a9]x!E+Oj b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@=a#`"FEi+h*NMq3p-Q2V + O U y   ; ) M x  ' K : ^   % I ]   C # G   !EBf(Ce'K Jn 0m/STx:w9]]C b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d=z7t1n+_h&:c p]  W   Q   K  E bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  SS)?*?44>'?'IIS T ^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33## { {""$k%k'')S*S,,.;/;113#4#668 9 ;;==?F@FBBD.E.GGI7J7LLNOQQSSUqVqXXZU[U]]_9`9bbdeggijlslsnnpcqcssuGvGxxz+{+}}ee׉׋II--  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSʃSʃSSdSSlSi!Si!S$S$S :S@ :So~So~SISISᐄS@ᐄS_S _SSSSSS SYS YSSd SCS CS1RS 1RSm[S m[SSS翅S翅SSSSS%S %S/S/SSpSAlStAlSySTySSSeS$eS S(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 4= bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 4= bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 4=  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` D=aTx4UyDh$4Xmx$Hh5 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@.R=a#$Hb$H> 4Gk -*j ,PwOsm 5cr4X W {   = z  < `   " _  ! E   Dh*g)MLp2o1UTx:w9]\B b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d.=z7\t31 n+h%b\  V   P   J  D bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  PP)@*@44>?IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``PP @ @  00   x x""$h%h'')X*X,,.H/H113848668(9(;;=>@@BCEEGGIPJPLLN|@82\ < `  |  @  \   <`|@\ <`~ Dd*Jn0Tt:Z~ @d  &!J!!!j"" #0###P$t$$%%%6&Z&&&z''(@( b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d8|xjtCplhd`\X V T R P NLJHFD bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  MM)7*744>?IIST^z^zhhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&rr\ \""$F%F'')0*0,,./11346y6y88:c;c==?K@KBBD1E1GGIJLLNNPpQpSSUVVVXXZ<[<]]_"`"bbd e g}g}iikclcnnpIqIssu/v/xxzz|h}hہNNTTƉƋ88  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSʃSʃSSdSSlSi!Si!S$S$S :S@ :So~So~SISISᐄS@ᐄS_S _SSSSSS SYS YSSd SCS CSRS RS\S \SKSKS翅S翅SSSSS%S %SSSSpS)lSt)lSySTySSS߉S$߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 3= bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 3= bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1=q  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 3=  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`!EPtv48\Uy DDh,#G|x6W{ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@=a#`"FEi+h*NMq3p2VY} U y O s   ; E i x  ; _ ; _   ! 1 U ^   D ' K   ACg)7f(LKo 1n0TSw9v8\[A b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d=z7t1na+8h%bf=\  V   P   J  D bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  LL)8*844>?IIS T ^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``LL 8 8  $$rr^ ^""$J%J'')6*6,,."/"11346688:p;p==?\@\BBDHEHGGIJLLNNPoQoSSUxVxXXZ`[`]]_H`Hbbd0e0ggijllnoq|q|ssudvdxxzL{L}}44xx``  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSS)˃S)˃SSdSSlSi!Si!S$S$S:S@:SSSISISᐄS@ᐄS_S _SSS}S}SS SYS YSSd SCS CS3SS 3SSo\S o\SSS'ȅS'ȅSSSSS'S 'SSSSpSlStlSySTySu Su SE߉S$E߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]) b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 MS  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]) b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 MS  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1])  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 MS   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`@[*N>Y}x<W{C:Uym9 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@_e kqw}$* 0, 6  &    <  y   B s   $ H m   * N g  0Ta6Z[<`>b@dBfDhFjHlJnLpNrPtRvTxVzX|Z~\  b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d_>|[u:P)xW[6  p  M  *   d A{X bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  KK)1*144>?IISS]l^lhhrNsN} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&nnT T""$:%:'') * ,,./1y1y335_6_88:E;E==?)@)BBD E G|G|IIK^L^NNP@Q@SSU"V"XXZ[]u]u__a[b[ddf=g=iiklnnpqsrsruuwTxTzz|}hhjjڋڍJJ  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lSSʃSʃSSLSSTSu!Su!S$S$S:S(:Su~Su~SUSUS퐄S(퐄SkSkSSSSSS SeS eSSL S+CS +CS[RS [RS[S [SߚSߚS=S=SUSUSSDSSS/lS /lSSSSS#S0#SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 = bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 = bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 =  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`=aCW{9aMq /Cg;%9]|;_ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@%IC=7 1+%~x bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1%~W0  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  NN)<*<44>*?*IIST^^hhrsss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``NN < <  **}}k k""$Y%Y'')G*G,,.5/5113#4#6689;;==?v@vBBDdEdGGIRJRLLNb#Ae`"FTx*W{ -j@m b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@+O!E; 1'~tj ` bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1+Y0 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  RR)<*<44>'?'IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%%q q""$[%['')E*E,,.///11346689;x;x==?b@bBBDLELGGI7J7LLNOQQSTV{V{XXZc[c]]_K`Kbbd3e3ggijllnoqwqwssu_v_xxzG{G}}//wwaa b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lSSʃSʃSSLSSTSu!Su!S$S$S :S( :So~So~SUSUS퐄S(퐄SkSkSSSSSS SeS eSSL S+CS +CSRS RS9\S 9\SSSSSSS SD SSSlS lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 = bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 = bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 =  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`=a&J~ . D{h,]`$AeHl @d b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@r l f  `  Z ~ T x NrHl bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1b;mF } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  MM)7*744>!?!IIST^z^zhhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%ss_ _""$J%J'')4*4,,./11346}6}88:g;g==?Q@QBBD;E;GGI%J%LLN O Q~Q~SSUdVdXXZJ[J]]_0`0bbdeggiikolonnpUqUssu;v;xxz!{!}}zz``ӉӋFF,, b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lSS#˃S#˃SSLSSTS!S!S$S$S:S(:S~S~SaSaSS(SwSwSSSwSwSS SqS qSSL S7CS 7CSRS RSO[S O[SSSնSնSSS SD SSSlS lS S S߉S0߉SSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 X bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 X bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 X  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1``A]Cd_ !Eha#Glc b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@VzL p B f 8 \ . R $ H >4 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1^5 c: } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  NN)8*844>"?"IIST^y^yhhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''ss] ]""$G%G'')1*1,,./11346z6z88:d;d==?N@NBBD8E8GGI"J"LLNOQ{Q{SSUaVaXXZG[G]]_-`-bbdeggiikglgnnpKqKssu/v/xxz{}}iiۄۆOO‰‹55 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLSSTS!S!S%S%S:S(:S{~S{~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CSQS QS[S [S[S[S[S[SUSUSSDSSS/lS /lS5S5SS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 )?  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 )?  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 )?   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`@cW{6;_Mq3 -Dh{ #Os:^z b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@><:8642 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1wR- } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  TT)@*@44>*?*IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddXX L L  @@44$$  ""$r%r'')^*^,,.J/J113646668"9";;=>@@BBDjEjGGIRJRLLN:O:QQS"T"VVX Y [~[~]]_f`fbbdNeNggi6j6llnoqqstvzvzxxzb{b}}JJ22rr b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS˃S˃SSLSSTS!S!S%S%S:S(:S~S~SySySS(SSSSSkSkSS SS SSL SOCS OCSSS SSW\S W\SSSSSCSCS#SD#SSS5lS 5lSMSMSS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xE b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 bS. bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xE b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 bS. bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xE  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 bS.  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`=a6Z-Q$H~?"Fl -4j2X|  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@B4&{m _QuCg5Y bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1sHr } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>1?1IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIYJYLLNGOGQQS5T5VVX#Y#[[]^``bbdvevggidjdllnMoMqqs1t1vvxy{{}}kk݂݄OO33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLSSTS!S!S%S%S:S(:Su~Su~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CS-SS -SSi\S i\SSSGSGSSSSDSSSlS lS S S߉S0߉S;S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 (> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 (> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 (>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`'K5(Ls+O=t>Eie /_f 0 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ 0 .  ,  *  (  & $"} bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1c>tO } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP)8*844>?IIST^t^thhrXsX} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$l l""$T%T'')<*<,,.$/$113 4 6688:h;h==?N@NBBD2E2GGIJLLNNPlQlSSUPVPXXZ4[4]]_`bbddfngniikRlRnnp6q6ssuvxxzz|p}pTTƄƆ88 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS˃S˃SSLSSTS!S!S%S%S:S(:S~S~SySySS(SSS S SqSqSS SS SSL SOCS OCSQS QSZS ZSSSSSiSiSSDSSS#lS #lS S S߉S0߉SkS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yG b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 cS0 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yG b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 cS0 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yG  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 cS0  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`=a:Os$ 5Yao%I?Uy %;_t b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@'K = / ! v  h  Z ~ Lp bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1GqF } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  MM)7*744>#?#IIS T ^^hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccMM 7 7  !!  jjT T""$>%>'')(*(,,./11335q6q88:[;[==?E@EBBD/E/GGIJLLNOQzQzSSUdVdXXZN[N]]_8`8bbd"e"ggi j llnnpeqessuIvIxxz-{-}}ggىًKK// b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃS SL SSTS!S!S)%S)%S:S(:S{~S{~SSS#S(#SSSSSSSS SS SSL SaCS aCSQS QS[S [SgSgSsSsS[S[S/SD/SSSAlS AlSSSYS0YSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  U@ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  U@ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  U@  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`3Cgwx7eFjz8SwIm|Ae; b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@79;=?AC!E bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1pM* } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  VV)H*H44>:?:IIS&T&^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddVV H H  ::,,  "#%{%{'')m*m,,._/_113Q4Q668C9C;;=5>5@@B'C'EEGHJJL M OOQQSnTnVVX\Y\[[]J^J``b6c6eeg h jjl m o}o}qqsctcvvxIyI{{}/~/mmߌߎQQÑÓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S˃S˃S{SL{SSTS!S!S %S %S:S(:S~S~SmSmSS(SSSSSeSeSS S}S }SSL SCCS CCSsRS sRS[S [SSS׮S׮S1S1S)SD)SSS;lS ;lS_S_S/S0/SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m: b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 bx" bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m: b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 bx" bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m:  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 bx"  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`=a+Oj5Imz:LpeB2V3Nr b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@5#vdRv@d.R@ . bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1d7 ] } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  ]])I*I44>5?5IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ii]] Q Q  EE99--!!  "#%%'(*w*w,,.c/c113O4O668;9;;;='>'@@BCEEGGIuJuLLN]O]QQSETEVVX-Y-[[]^``bbdqeqggiYjYllnAoAqqs)t)vvxy{{}}mmUUɇɉ==%% b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTS!S!S)%S)%S:S(:S{~S{~SSS#S(#SSSSSSSS SS SSL SaCS aCSRS RS[S [SESESSSSSSDSSS/lS /lSSS߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  T> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  T> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  T>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` Dat 06ZPt  -<`u<Imm)Im b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@#G% I ' K ) M + O - Q /S1U bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1dA{X } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  RR)@*@44>.?.IIST^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddRR @ @  ..  o o""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGIVJVLLN@O@QQS*T*VVXY[[]]_s`sbbdYeYggi?j?lln#o#qqstvvvvxxzX{X}}::ppRR b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S˃S˃S{SL{SSTS!S!S%S%S:S(:S~S~SsSsS S( SSSSSkSkSS SS SSL SICS ICS)QS )QSeZS eZSSSKSKSQSQSSDSSS#lS #lS S S߉S0߉S_S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n< b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 cy$ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n< b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 cy$ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n<  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 cy$  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`=a />bm#,`7[m"F%7[f% b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@o ] K o 9 ] ' K  9 'z bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations10V)| } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  WW)?*?44>'?'IIS T ^}^}hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggWW G G  77''  ""$k%k'')S*S,,.;/;113#4#668 9 ;;==?g@gBBDOEOGGI7J7LLNOQQSSUqVqXXZU[U]]_9`9bbdeggijlslsnnpWqWssu;v;xxz{}}uuYYˉˋ==!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTS{!S{!S$S$S:S(:S~S~S[S[SS(SqSqSSSSSS SkS kSSL S1CS 1CSQS QSZS ZSSSSS[S[S5SD5S S SGlS GlSSSqS0qSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 =  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 =  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 =   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`&-QpcFj%IPt@ 6Hls_ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ .4:@"F(L.R4X bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 iH' } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  TT)D*D44>4?4IIS T ^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$ | |""$l%l'')\*\,,.L/L113<4<668,9,;;=>@@B C EEGGItJtLLN`O`QQSLTLVVX8Y8[[]$^$``bceeggirjrlln^o^qqsJtJvvx6y6{{}"~"pp\\ґғ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S ˃S ˃SoSLoSSTS{!S{!S$S$S:S(:S~S~S[S[SS(SqSqSSS_S_SS SkS kSSL S1CS 1CSQS QS[S [SOSOSSSSSSDSSS)lS )lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b/ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 KW bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b/ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 KW bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b/  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 KW  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` DvTx 6X|Fj(:^8\}@*No b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@(ycMq7[!E /j bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1U&wH } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  QQ);*;44>#?#IIS T ^|^|hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **vv` `""$J%J'')4*4,,./11346}6}88:g;g==?O@OBBD5E5GGIJLLNOQtQtSSUZVZXXZ@[@]]_&`&bbd e ggiikelennpKqKssu1v1xxz{}}ppVVȉȋ:: b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃS SL SSTS{!S{!S$S$S:S(:S~S~S[S[SS(SqSqSSSSSS SkS kSSL S1CS 1CS7RS 7RSs[S s[SSSSSSS#SD#SSS5lS 5lS)S)S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 =  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 =  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 =   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`;gb )M#GnLp/Sz2$;_ 1 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@:^@ d F j L p R v X | _e  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1eD#a } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP)<*<44>(?(IIST^^hhrlsl} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((vvb b""$N%N''):*:,,.&/&11346688:t;t==?`@`BBDLELGGI8J8LLN O QQSTV|V|XXZd[d]]_L`Lbbd4e4ggijllnoqxqxssu`v`xxzH{H}}00tt\\ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S˃S˃SuSLuSSTSu!Su!S$S$S:S(:S~S~SUSUS퐄S(퐄SkSkSSS_S_SS SeS eSSL S+CS +CSQS QSZS ZS=S=SSS3S3SSDSSSlS lS S SW߉S0W߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c1 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 JX bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c1 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 JX bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c1  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 JX  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$+OqZ~SwA`5Y.RRv;  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Tx>b( L  6 q [  E i /S bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1j; ] } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  OO)5*544>?IISS]l^lhhrNsN} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **rrX X""$>%>'')$*$,,. / 1}1}335c6c88:I;I==?-@-BBDEGGIIKbLbNNPDQDSSU&V&XXZ[]y]y__a[b[ddf=g=iiklnnpqsrsruuwTxTzz|6}6jjڋڍJJ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSSLSSTS!S!S$S$S#:S(#:S~S~SaSaSS(SwSwSSSSSS SqS qSSL S7CS 7CSGQS GQSZS ZS˙S˙SSSaSaS/SD/SSSAlS AlSeSeS5S05SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 >  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 >  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 >   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$Hkl0T BffSw<<`94 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@& 0: D*N4X>bHl bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1cD% } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  RR)@*@44>.?.IIST^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddRR @ @  ..  o o""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGIVJVLLN@O@QQS*T*VVXY[[]]_s`sbbd]e]ggiGjGlln1o1qqstvvxy{z{z}}ddقلNNÇÉ88"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS˃S˃SiSLiSSTS!S!S$S$S}:S(}:S~S~SaSaSS(SwSwSSSYSYSS SqS qSSL S7CS 7CSQS QS[S [SgSgSSS=S=S)SD)SSS;lS ;lS5S5SS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W$ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 N o bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W$ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 N o bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W$  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 N o  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`9>b',PjzRv>h:^,Vz b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$|tl d\TxLpDh bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1X0` } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  VV)@*@44>+?+IIST^^hhrqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiYY I I  99)) u u""$_%_'')I*I,,.3/311346689;|;|==?f@fBBDPEPGGI;J;LLN#O#QQS T VVXXZg[g]]_O`Obbd7e7ggijllnoq{q{ssucvcxxzK{K}}33{{ee b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSSLSSTS!S!S$S$S#:S(#:S~S~SaSaSS(SwSwSSSSSS SqS qSSL S7CS 7CSRS RSC[S C[SSSSSSSSDSSS/lS /lS S S߉S0߉SwS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 > bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 > bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 >  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`2fPt(L'Rv)KoTx 1+Vz b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Rv\  f  p  z # -7A bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1gH) k } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  QQ);*;44>%?%IIS T ^~^~hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))wwc c""$N%N'')8*8,,."/"113 4 6688:k;k==?U@UBBD?E?GGI)J)LLNOQQSSUhVhXXZN[N]]_4`4bbdeggijlslsnnpYqYssu?v?xxz%{%}} ~~dd׉׋JJ00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS ˃S ˃SoSLoSSTS!S!S$S$S}:S(}:S~S~SaSaSS(SwSwSSSYSYSS SqS qSSL S7CS 7CSQS QS [S  [SUSUS۶S۶SuSuSSDSSS/lS /lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X& b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 M o bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X& b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 M o bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X&  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 M o  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$Hid;__Hl 11Uv,P' b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@<`4 X , P $ H  @  8 0( bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations18h@ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  RR)<*<44>&?&IIS T ^}^}hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++wwa a""$K%K'')5*5,,./113 4 6~6~88:h;h==?R@RBBD.?.IIST^^hhrrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hh\\ P P  DD88((  "#%v%v'')b*b,,.N/N113:4:668&9&;;=>@@BBDnEnGGIVJVLLN>O>QQS&T&VVXY[[]]_j`jbbdReRggi:j:lln"o"qqs t v~v~xxzf{f}}NN‚„66vv b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃScSLcSSTS!S!S%S%Sw:S(w:S~S~SgSgSS(S}S}S梅S梅SSSSSS SwS wSSL S=CS =CSQS QS[S [SISISǯSǯSmSmS5SD5S S SGlS GlS_S_S/S0/SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Qb bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Qb bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Qb  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` ..Rx <Z~i<`&SwwB=aa b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$2@*N8\FjTxb  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1cF)  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  WW)E*E44>5?5IIS#T#^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiWW E E  33!!t t""$b%b'')P*P,,.>/>113,4,6689;;=>@@BBDmEmGGI]J]LLNKOKQQS9T9VVX'Y'[[]^``bcezezggihjhllnQoQqqs5t5vvxy{{}}ooSSŇʼn77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃSSLSSTS!S!S%S%S):S():S~S~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CSRS RSC[S C[SSSSSSSSDSSS)lS )lS S S߉S0߉SMS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 = bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 = bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 =  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`=a)>bx1UDhm%IJns>Pty b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@; _ Q u g  }  , B4XJn bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1jQ8 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  TT)<*<44>"?"IIST^x^xhhr\s\} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhXX H H  88((p p""$X%X'')@*@,,.(/(11346688:l;l==?R@RBBD6E6GGIJLLNNPpQpSSUTVTXXZ8[8]]_`bbdegrgriikVlVnnp:q:ssuvxxz{}t}tXXʄʆ<< b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃScSLcSSTS!S!S%S%Sw:S(w:S~S~SgSgSS(S}S}S梅S梅SSSSSS SwS wSSL S=CS =CSQS QS[S [SgSgS7S7SSS#SD#SSS5lS 5lSSS߉S0߉S}S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 P c bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 P c bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 P c  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$Hr,hVzzPt:^^8\!EEi b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@p ~  )  7 ! E / S =aKoY} bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1oR5{ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  QQ);*;44>'?'IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggQQ ; ;  %%nnX X""$B%B''),*,,,./11346u6u88:_;_==?I@IBBD3E3GGIJLLN O Q~Q~SSUhVhXXZR[R]]_<`M?MIIS9T9^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1yykk ] ]  OOAA33%%  "#%%' ( **,,.t/t113f4f668X9X;;=J>J@@B@?@IIS(T(^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvjj ^ ^  RRFF::..  ""#"%%'(**,,.p/p113\4\668H9H;;=4>4@@B C EEG H JJLLNhOhQQSPTPVVX8Y8[[] ^ ``bce|e|ggidjdllnLoLqqs4t4vvxy{{}~xx``ԇԉHH00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS˃S˃SiSiSS$S}:S}:S~S~S[S$[SSSYSYSS$SQSQSZS4ZSST S_ S _ SAS ASS SSlS SlS;S ;S S SS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations16jS b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations16jS b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations16jS b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 =a<kOs  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ww  ee)S*S44>???IIS'T'^^h i sysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1wwee S S  AA//   ""$p%p'')^*^,,.L/L113:4:668(9(;;=>@@BCE{E{GGIgJgLLNQOQQQS;T;VVX%Y%[[]^``bbdjejggiPjPlln4o4qqstvvxxzi{i}}KK--aa b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃS]S ]SSSq:Sq:S~S~SUSUS驄Sp驄SMSMSSSQS`QS[S[SS$ S_ S _ S5S 5S S  SGlS GlS S  S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1{} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1{} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1{ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 u W{;_A  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10vv  ff)N*N44>4?4IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvff V V  FF66&&  "#%z%z'')b*b,,.J/J1132426689;;=>@v@vBBD^E^GGIDJDLLN(O(QQS T V~V~XXZb[b]]_F`Fbbd*e*ggijllnnpdqdssuHvHxxz,{,}}ff؉؋JJ.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS'S 'SSS;:S;:S~S~SmSmSSpSSSSSwQS`wQSZSZSS$ S_ S _ SSS SS+S +SelS elSS SSS1S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Bf2w Y};_  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10uu  ee)U*U44>C?CIIS/T/^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1uuee U U  EE55%%  "#%}%}'')m*m,,.]/]113M4M668=9=;;=->-@@BCEEG H JJLLNoOoQQS[T[VVXGYG[[]3^3``bceeg h jjllnmomqqsYtYvvxEyE{{}1~1 kk b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃScS cSSSw:Sw:S~S~S7S7S梅Sp梅SSSSSSSQS`QSYZSYZSS$ S_ S _ S)S )SS S;lS ;lS S  S߉S߉SqS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1b} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1b} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1b k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 =a#`"F  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10oo  ZZ)D*D44>*?*IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1oo[[ G G  33  i i""$S%S'')=*=,,.'/'11346688:p;p==?V@VBBD mOs  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  __)K*K44>5?5IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ss__ K K  77##q q""$]%]'')I*I,,.5/5113!4!668 9 ;;==?o@oBBD[E[GGIEJELLN-O-QQSTVVXXZq[q]]_Y`YbbdAeAggi)j)llnoqqssumvmxxzU{U}}==%% ii b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃScS cSSSw:Sw:S~S~S=S=S梅Sp梅SSSSSSSQS`QSYZSYZSS$ S_ S _ SS SS S)lS )lS S  Si߉Si߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1c} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1c} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1c k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 =a#`"F  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  ZZ)@*@44>"?"IIST^u^uhhrUsU} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qq]] I I  55!!  }}c c""$I%I'')/*/,,./11335n6n88:T;T==?6@6BBDEGGIIKkLkNNPMQMSSU/V/XXZ[]]__adbdddfFgFiik(l(nnp q s{s{uuw]x]zz|?}?!!qqQQ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS!S !SSS5:S5:S~S~S[S[SSpSSSSSqQS`qQSZSZSS$ SY S Y SAS ASS SSlS SlSwS wSGSGSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 GkA^;_  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  __)M*M44>9?9IIS#T#^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qq__ M M  ;;)) | |""$j%j'')X*X,,.F/F113444668"9";;=>@@BBDuEuGGIaJaLLNKOKQQS5T5VVXY[[] ^ `~`~bbdhehggiRjRlln.?.IIST^^hhrrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nn^^ N N  >>.. z z""$d%d'')N*N,,.8/8113"4"668 9 ;;==?k@kBBDUEUGGI>J>LLN&O&QQSTVVXXZj[j]]_R`Rbbd:e:ggi"j"lln o q~q~ssufvfxxzN{N}}66||ff b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS!S !SSS/:S/:S~S~S[S[SSpS S SSSQS`QSZSZSS$ SY S Y S)S )SS S;lS ;lS S  S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 =a<kMq  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  \\)F*F44>.?.IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp\\ H H  44    n n""$Y%Y'')C*C,,.-/-11346689;v;v==?`@`BBDJEJGGI2J2LLNOQQSSUqVqXXZW[W]]_=`=bbd#e#ggi j l|l|nnpbqbssuHvHxxz.{.}}mmSSƎƐ99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS˃S˃SiS iSSS}:S}:S~S~SUSUSSpSYSYSSSGQS`GQSZSZSS$ SY S Y S#S #SS S5lS 5lS S  S߉S߉SqS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 =a /r@d2  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  YY)C*C44>+?+IIST^^hhrfsf} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nnZZ F F  22  ~~h h""$R%R'')<*<,,.&/&11346688:o;o==?Y@YBBDCECGGI+J+LLNOQQSSUjVjXXZP[P]]_6`6bbdeggiikplpnnpTqTssu8v8xxz{}}rrVVɉɋ<<"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃSS SSS/:S/:S~S~SySySSpS S SSSkQS`kQSZSZSS$ SS S S S/S /SS SAlS AlSMS MSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1*} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1*} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1* k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 h4X&h6Z  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  aa)M*M44>5?5IIST^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qqee Y Y  MMAA11  " # %%'')k*k,,.W/W113C4C668/9/;;=>@@BCEuEuGGI]J]LLNEOEQQS-T-VVXY[[]]_q`qbbdYeYggiAjAlln)o)qqstvvxxzm{m}}UUɂɄ==""{{ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS ˃S ˃SoS oSSS:S:S~S~S%S%SSpS_S_SSSQS`QSSZSSZSS$ SS S S S/S /SS SAlS AlS_S _S/S/SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[( b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1S} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[( b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1S} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[( b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1S k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 =a#`"F  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  XX)F*F44>4?4IIS"T"^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjXX F F  44""u u""$c%c'')Q*Q,,.?/?113-4-6689;;= > @@BBDnEnGGI\J\LLNJOJQQS8T8VVX&Y&[[]^``bceyeyggigjgllnPoPqqs4t4vvxy{{}}nnRRćĉ66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃSS SSS):S):S~S~SySySSpSSSSSAQS`AQS}ZS}ZSS$ SS S S SS SS S)lS )lS S  S߉S߉SMS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1+} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1+} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1+ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 j,POs5  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10oo  [[)C*C44>'?'IIS T ^}^}hhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1oo__ O O  ??// w w""$_%_'')G*G,,.///11346688:s;s==?W@WBBD;E;GGIJLLNOQuQuSSUYVYXXZ=[=]]_!`!bbdegwgwiik[l[nnp?q?ssu#v#xxz{}y}y]]τφ??!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS ˃S ˃SoS oSSS:S:S~S~S%S%SSpS_S_SSSQS`QS[S[SS$ SS S S S#S #SS S5lS 5lSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z( b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1S} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z( b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1S} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z( b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1S k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 =a8\2V,P  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  TT)>*>44>(?(IIST^^hhrhsh} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjTT > >  ((qq[ [""$E%E'')/*/,,./11346x6x88:b;b==?L@LBBD6E6GGI J LLN O QQSSUiViXXZS[S]]_=`=bbd'e'ggijllnnpjqjssuNvNxxz2{2}}llމދPPŽ44 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃSS SSS):S):S~S~S[S[SSpSSSSS5QS`5QSqZSqZSS$ SS S S S;S ;SS SMlS MlSS SeSeS S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 b$H Gk -  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  __)Q*Q44>A?AIIS-T-^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mm__ Q Q  CC55''  " # %%'')v*v,,.h/h113Z4Z668L9L;;=>>>@@B0C0EEG"H"JJLMOOQQSuTuVVXcYc[[]Q^Q``b=c=eeg'h'jjlmooqqsjtjvvxPyP{{}6~6rrVVȑȓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S˃S˃SuS uSSS:S:S~S~SISISSpS_S_SSS;QS`;QSwZSwZSS$ SS S S S/S /SS SAlS AlSeS eS5S5SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d1 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d1 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d1 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Dh4vBf2  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  ^^)J*J44>4?4IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jj^^ R R  FF::..""  "#%%'(*x*x,,.d/d113P4P668<9<;;=(>(@@BCEEGHJtJtLLN\O\QQSDTDVVX,Y,[[]^``bbdpepggiXjXlln@o@qqs(t(vvxy{{}}llTTȇȉ<<$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃSS SSS#:S#:S~S~S[S[SSpSSSSSQS`QSZSZSS$ SS S S S#S #SS S5lS 5lS#S #S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 w Vz4X6  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  YY)G*G44>3?3IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kkYY G G  55##v v""$d%d'')R*R,,.@/@113.4.6689;;= > @@BBDoEoGGI[J[LLNEOEQQS/T/VVXY[[]^`x`xbbd^e^ggiDjDlln(o(qqs t v{v{xxz]{]}}??!!ssUU b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S˃S˃SuS uSSS:S:S~S~SISISSpS_S_SSSQS`QS[S[SS$ SS S S SS SS S)lS )lS S  S߉S߉SeS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c1 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c1 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c1 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 =a;_9]7[  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  ZZ)B*B44>(?(IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjZZ J J  ::**   ""$n%n'')V*V,,.>/>113&4&6689;;==?j@jBBDRERGGI8J8LLNOQQSTVrVrXXZV[V]]_:`:bbdeggijltltnnpXqXssu>"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃSS SSS:S:S~S~SmSmSSpSSSSSMQS`MQSZSZSS$ SM S M S5S 5S S  SGlS GlSS SkSkSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1'} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1'} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1' k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 d,PX| D  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  YY)I*I44>7?7IIS#T#^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiYY I I  99))   ""$q%q'')a*a,,.Q/Q113A4A668191;;=!>!@@BCEEGHJwJwLLNcOcQQSOTOVVX;Y;[[]'^'``bceeggiujullnaoaqqsMtMvvx9y9{{}%~%ss__ՑՓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S˃S˃S{S {SSS:S:S~S~SSSSpSeSeSSSRS`RS[S[SS$ SM S M S S  SS SlS lS S  S߉S߉SYS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m: b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1S} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m: b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1S} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m: b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1S k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 @8\Rvl  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  NN)8*844>?IIST^w^whhr[s[} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''ss] ]""$G%G'')1*1,,./11346z6z88:d;d==?J@JBBD0E0GGIJLLNNPoQoSSUUVUXXZ;[;]]_!`!bbdegzgziik`l`nnpFqFssu,v,xxz{}}kkބކOO33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃS S  SSS:S:S~S~SmSmSSpSSSSS1RS`1RSm[Sm[SS$ SM S M SS SS S/lS /lS#S #S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1(} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1(} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1( k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ~#%')  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  SS)?*?44>)?)IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++yye e""$Q%Q'')=*=,,.)/)11346689;w;w==?c@cBBDOEOGGI9J9LLN!O!QQS T V}V}XXZe[e]]_M`Mbbd5e5ggijllnoqyqyssuavaxxzI{I}}11uu]] b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S˃S˃S{S {SSS:S:S~S~SSSSpSeSeSSSQS`QSZSZSS$ SM S M SS SS S lS  lS{ S { SK߉SK߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l: b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1S} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l: b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1S} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l: b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1S k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 rVz:^B  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  NN)4*444>?IISS]i^ihhrIsI} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))qqW W""$=%='')#*#,,. / 1|1|335b6b88:H;H==?*@*BBD E G}G}IIK_L_NNPAQASSU#V#XXZ[]v]v__aXbXddf:g:iiklnnpprosouuwQxQzz|3}3eeՋՍEE b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYSʃSʃS S  SSS:S:S{~S{~SOSOSSpSSSSSQS`QSZSZSS$ SG S G S#S #SS S5lS 5lS_S _S/S/SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 mCg=Z~  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>-?-IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIUJULLN?O?QQS)T)VVXY[[]]_r`rbbd\e\ggiFjFlln2o2qqstvvxy{{{{}}eeڂڄOOćĉ99## b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYS˃S˃S{S {SSS:S:S~S~SCSCSSpSkSkSSSQS`QS[S[SS SG S| G SS SE~S E~S#NS #NS)S )SމSމSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vC b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1}} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vC b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1}} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vC b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1} k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations14!E1U  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  OO)9*944>"?"IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbRR B B  22""n n""$X%X'')B*B,,.,/,11346689;u;u==?_@_BBDIEIGGI2J2LLNOQQSTVvVvXXZ^[^]]_F`Fbbd8e8ggi j llnoq|q|ssudvdxxzL{L}}44zzdd b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYSʃSʃSS SSS:S:S{~S{~SOSOSSpSSSSS[RS`[RS[S[SS$ SG S G SS SS S#lS #lS S  S߉S߉SkS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 *8$H2V  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP):*:44>"?"IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((vvb b""$M%M'')7*7,,.!/!113 4 6688:j;j==?T@TBBD>E>GGI&J&LLN O QQSSUeVeXXZK[K]]_1`1bbdeggiikplpnnpZqZssu@v@xxz&{&}} ee؉؋KK11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYS˃S˃S{S {SSS:S:S~S~S=S=SSpSkSkSSS_QS`_QSZSZSS SG S| G SS S9~S 9~SNS NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uC b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1|} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uC b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1|} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uC b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1| k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1q't Vz8\  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  MM)7*744>?IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&rr\ \""$F%F'')0*0,,./11346y6y88:c;c==?M@MBBD7E7GGIJLLNOQxQxSSU^V^XXZD[D]]_*`*bbdeggiiklllnnpPqPssu4v4xxz{}}nnRRʼnŋ88 b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LSSʃSʃSS SSS:S:Su~Su~SaSaSSpSSSSSCRS`CRS[S[SS SG S| G S#S #SK~S K~S)NS )NS;S ;S1މS1މSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1$} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1$} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1$ k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1r,4X` D  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  UU)C*C44>+?+IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeYY M M  AA55%%  ""$u%u'')a*a,,.M/M113949668%9%;;=>@@BBDkEkGGISJSLLN;O;QQS#T#VVX Y [[]]_g`gbbd[e[ggiCjClln+o+qqstvvxxzo{o}}WW˂˄??$$ }} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LSS˃S˃SS SSS:S:S~S~SSS Sp SqSqSSSQS`QSZSZSS SG S| G S)S )SQ~S Q~S/NS /NSYS YSOމSOމSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1S} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1S} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1S k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1&(*  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10^^  LL)D*D44>2?2IIS T ^^hiss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1^^LL : :  (({{i i""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIZJZLLNHOHQQS6T6VVX$Y$[[]^``bcewewggiejellnZoZqqs>t>vvx"y"{{}~xx\\·Ή@@$$ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LSSʃSʃSS SSS :S :So~So~SgSgSSpSSSSSYQS`YQSZSZSS SG S| G S S  S3~S 3~SNS NS S  S݉S݉SgS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1%} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1%} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1>r b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1% k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1Sw=a'K6Z  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  OO);*;44>?IIST^^hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33##k k""$W%W'')?*?,,.'/'11346688:k;k==?O@OBBD3E3GGIJLLNNPmQmSSUQVQXXZ5[5]]_`bbd e g{g{iik_l_nnpCqCssu'v'xxz { }}}}aaӄӆCC%% b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LSS˃S˃SS SSS:S:S~S~SSS Sp SqSqSSSPS`PS/ZS/ZSS SG S| G SS S?~S ?~SNS NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~L b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1S} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~L b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1S} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~L b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1S k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1x Jn 0m  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10^^  HH):*:44>$?$IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1^^HH 2 2  {{eeO O""$A%A'')+*+,,./11335t6t88:^;^==?H@HBBD2E2GGIJLLNOQ{Q{SSUeVeXXZO[O]]_9`9bbd#e#ggi j llnoqrqrssuVvVxxz:{:}}ttXXʎʐ<< b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations1MSoʃSoʃSS|SsSsSSS"S|"S&SH&S9S9SE~SPE~S냄SX냄SiSiSSPSSSYSYSS$ ScS, cSS SyS$ ySS S?DS| ?DScSSXcSS\S|\S盅S盅S%ąSx%ąSS\SaSaS_&S_&SS8SSS/lS/lSySyS)Sl)S߉S߉SS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ps b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1/ bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ps b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1/  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ps  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1/ bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`<`Mq/S8\ 7FjBfhPt"Lp b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Mq/S5dFj(L .{]?c! E   ' t  V z  8 \   >   oUySw;_Ei!E7[+)Mz?` 1Gk#-Qx7jlRv8\B(w] b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dMg4Nh5Q  Z  o /  >   Y \ +  zI bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10==  LL)@*@44>+?+IIST^^his}s}} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1==--    //  k k""$_%_'')I*I,,.3/311346689;|;|==?f@fBBDPEPGGI;J;LLN#O#QQS T VVXXZg[g]]_O`ObbdCeCggi+j+llnoqqssuovoxxzW{W}}??''qq# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xMS/˃S/˃SSdSSlS]!S]!S$S$S:S@:S S S=S=SՐS@ՐSSS SSSSSSS SMS MSSd SCS CSRS RS!\S !\SiSiSeSeS!S!S5S5S3%S 3%S#S#SSpS5lSt5lSySTyS_S_S/S$/SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d0 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ?!~ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d0 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ?!~ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d0  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ?!~  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`>Ko.R)Mg+bEiv  #GTx0T b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@?c)h.RW{AFj 0o5Y^$ H    O s   9 x  > b   ( g  -QVzAZ~FjTx 0Nro5YHlBf^$H<`6ZMq70Tv=a b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d?~=|;z9x7v5t7 3   r i  1 B p  /  n  -l bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  PP)@*@44>.?.IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$]%]'')K*K,,.9/9113'4'6689;;==?X@XBBDFEFGGIVJVLLN@O@QQS*T*VVXY[[]]_s`sbbd]e]ggiGjGlln=o=qqs't'vvxy{{}}ppZZχωDD..  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1MSiʃSiʃSS|SsSsSSS"S|"S&SH&S9S9SE~SPE~S냄SX냄SiSiSSPSSSYSYSS$ ScS, cSS SyS$ ySS S?DS| ?DS SSX SSE\S|E\SSSυSxυSS\SSS *S *SS8SSS#lS#lSySyS Sl S߉S߉S_S } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Op b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1/ bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Op b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1/  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Op  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1/ bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`>~Mq D /Smd5Os*1UdJn b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@~ Dd*Jn0Tt:Z ~   @ d    & J  j 0Pt6Zz@\ <`\|Nr@@d\2V $H<`:,|@\  &?&IIST^^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1<<((    ##ooY Y""$K%K'')5*5,,./113 4 6~6~88:h;h==?R@RBBD}Cg -l2V[!E Jn4s*N9]$HB<6 0*$ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d?~=|;z9x7v5t4  s  2 q 0  o t  . Mm&X bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  MM);*;44>%?%IIS T ^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%ss_ _""$N%N'')8*8,,."/"113 4 6688:k;k==?2@2BBDEGGI)J)LLNOQQSSUhVhXXZN[N]]_4`4bbdeggijlslsnnpeqessuKvKxxz1{1}}ppVVɎɐ<<  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSʃSʃSSdSSlSi!Si!S$S$S:S@:Su~Su~SISISᐄS@ᐄS_S _SSSSSS SYS YSSd SCS CSESS ESS\S \SɛSɛSSSSSSS%S %SSSSpS/lSt/lSySTySMSMSS$SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 0? bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 0? bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 0?  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`9]?c3W0T'K7[?.R5Y,P b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Dh8|Lp@Tx$H\,P d4 X   ( l  < `   0 t  F j  >X|+,P$!jx>bn 6|d PtZ~$HPtFjd8\ 0vJnB b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dDLT\ d(n 4 z i @ A    L o   X d bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  MM);*;44> ? IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__MM ; ;  ))||j j""$X%X'')F*F,,.4/4113"4"6689;;==?u@uBBDcEcGGI1J1LLNOQQS T VVXY[[] ^ ``bbdpepggi^j^llnQoQqqs5t5vvxy{{}}ooSSŇʼn77  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSS)˃S)˃SSdSSlSi!Si!S$S$S:S@:SSSISISᐄS@ᐄS_S _SSS}S}SS SYS YSSd SCS CSQS QSZS ZS%S%S%ͅS%ͅSSS7S7S5)S 5)SSSSpS)lSt)lSySTyS5S5SS$SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y% b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 JQ  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y% b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 JQ  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y%  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 JQ   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Mq?Qu/S);_z5%Id3fNr b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@q*0TZ~=Cgm&, P  V z     9  ? c  i"(L:8Tx 6~94Cg2 0q, .6Z,d)MW{@Jnx 3  =!a!!!k""#&###0$T$$$^%% b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dqbSD5&yj D   _ z  R U  E 0  8  +wj bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  VV)B*B44>,?,IIST^^hhrrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffZZ N N  BB66&&  ""$t%t'')`*`,,.L/L113848668$9$;;=>@@BBDlElGGITJTLLNb4y K o   A    X |  * N   c  1U#f4X&i7[)l:^,o=a /@dr@d6Z2,Pu"F<2(u b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dE OYc(m2u8 { >   D   F J  tPK" bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  II)3*344>>HIS T ^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__II 3 3  ||ffP P""$:%:'')$*$,,./11335m6m88:W;W==?A@ABBD+E+GGIIKiLiNNPSQSSSU`V`XXZJ[J]]_4`4bbdeggijl}l}nnpiqissuMvMxxz1{1}}kk݉݋OO33  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSS/˃S/˃SSdSSlSi!Si!S$S$S:S@:SSSISISᐄS@ᐄS_S _SSS}S}SS SYS YSSd SCS CSRS RS\S \S]S]S=DžS=DžSSS?S?S='S ='SSSSpSlStlSySTyS S S߉S$߉S5S(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]) b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 KQ  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]) b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 KQ  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1])  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 KQ   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`x"Fy89[LpMq  /S D!Eg b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@TxDh4X$H8(|l\Lp< `  , P   @   0   t  dRv>b*N:&xdPt<`(L8*$(v&b$Nr":^} &J{yw b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dT(|P$xL tH l >   b  4   fXA*|w bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP)8*844>?IIST^x^xhhr\s\} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$l l""$T%T'')<*<,,.$/$113 4 6688:h;h==?N@NBBD2E2GGIJLLNNPlQlSSUPVPXXZ4[4]]_`bbdegrgriikVlVnnp:q:ssuvxxzz|Q}QÁ55<<  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLSSTS{!S{!S$S$S:S(:S{~S{~S[S[SS(SqSqSSSSSS SkS kSSL S1CS 1CS9SS 9SSu\S u\SSS߾S߾SSSSDSSS/lS /lSSSSS#S0#SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ? bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ? bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ?  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`?c8\,Po+0T|7[$H(Lht@ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@d:^4X.R(L"F@: bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 v O (   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  WW)C*C44>/?/IIST^^hhrsss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccWW K K  ??33''  "#%%'')q*q,,.]/]113I4I668595;;=!>!@@B C EEGGIoJoLLNWOWQQS?T?VVX'Y'[[]^``bbdkekggiSjSlln;o;qqs#t#vvx y {{}}ggۂۄOOÇÉ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS#˃S#˃SSLSSTS{!S{!S$S$S:S(:S~S~S[S[SS(SqSqS S SqSqSS SkS kSSL S1CS 1CSQS QS [S  [SUSUSSS S S)SD)SSS;lS ;lSSSYS0YSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 U~ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 U~ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 U~  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`?c@W{36Mqv,DhCg7"9]y b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@OsQuSwUyW{Y}[] bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1O , f C  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  XX)J*J44>7@@B)C)EEGHJJL M OOQQSpTpVVX^Y^[[]L^L``b8c8eeg"h"jjl m ooqqsetevvxKyK{{}1~1ooSSőœ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLSSTSu!Su!S$S$S:S(:Su~Su~SUSUS퐄S(퐄SkSkSSSSSS SeS eSSL S+CS +CSQS QSZS ZS%S%SƅSƅS S SSDSSSlS lS S S߉S0߉SYS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ? bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ? bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ?  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`?c<Qu9&;_w%Ina3VzKo b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@4 .("{uoi  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 Z 3 e> } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  SS);*;44>#?#IIST^y^yhhr]s]} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33## { {""$g%g'')O*O,,.7/711346689;{;{==?c@cBBDKEKGGI3J3LLNOQQSSUmVmXXZQ[Q]]_5`5bbdeggiikolonnpSqSssu7v7xxz{}}qqUUljNj99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS#˃S#˃SSLSSTS!S!S$S$S:S(:S~S~SaSaSS(SwSwSSSwSwSS SqS qSSL S7CS 7CS-SS -SSi\S i\SSSDžSDžScScSSDSSS)lS )lSSS߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 V bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 V bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 V  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Uy'KCg*Nu1U?vC@g 1h2 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@_ace g ikm  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 }Z7 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  RR)@*@44>.?.IIST^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddRR @ @  ..  o o""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGIVJVLLN@O@QQS*T*VVXY[[]]_s`sbbdYeYggi?j?lln#o#qqstvvvvxxzX{X}}::ppRR b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃS SL SSTS!S!S%S%S:S(:S{~S{~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CS-SS -SSi\S i\SSS9S9SSS SD SSSlS lS S S߉S0߉SSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 %A  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 %A  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 %A   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`7[p (LZ~)MEit.R>u?f 0 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ywusqomk bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1(  ^ 9   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  KK)5*544>?IIST^v^vhhr\s\} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``LL 8 8  $$ppZ Z""$D%D'').*.,,./11346w6w88:a;a==?I@IBBD/E/GGIJLLNNPnQnSSUTVTXXZ:[:]]_ ` bbdegygyiik_l_nnpEqEssu+v+xxz{}}jj݄݆PP‰‹44 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS˃S˃SSLSSTS!S!S%S%S:S(:S~S~SySySS(SSSSSkSkSS SS SSL SOCS OCSQS QS%[S %[SmSmSSS%S%S5SD5S S SGlS GlSSSkS0kSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wC b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 `Q. bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wC b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 `Q. bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wC  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 `Q.  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`?cAY})M<Tx76Nr! 0Hlr b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@+ 17=C%I+O1U bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1Z 9  w V  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  VV)F*F44>6?6IIS"T"^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&& ~ ~""$n%n'')^*^,,.N/N113>4>668.9.;;=>@@BCEEGGIvJvLLNbObQQSNTNVVX:Y:[[]&^&``bceeggitjtlln`o`qqsLtLvvx8y8{{}$~$rr^^ԑԓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLSSTS!S!S%S%S:S(:S{~S{~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CSRS RS\S \ScScS%DžS%DžS-S-SSDSSS lS  lS{ S{ SK߉S0K߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 $@ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 $@ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 $@  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`;Sw'K+Cg~W{1UlCZ~C b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@i g eca_][ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 o J%[ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  KK)1*144>?IISS]h^hhhrJsJ} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&nnT T""$:%:'') * ,,./1y1y335_6_88:E;E==?)@)BBD E G|G|IIK^L^NNP@Q@SSU"V"XXZ[]u]u__aWbWddf9g9iiklnnpprnsnuuwPxPzz|2}2ff֋֍FF b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS˃S˃SSLSSTS!S!S%S%S:S(:S~S~SySySS(SSS S SqSqSS SS SSL SOCS OCS?SS ?SS{\S {\SÛSÛS-ȅS-ȅSSSSDSSS/lS /lS#S#S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yG b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 aQ0 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yG b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 aQ0 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yG  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 aQ0  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`?c(L+,Plx2V D|$Hp<t@ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@7[=aCgImOsUy[ a  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations15 sR1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP)<*<44>(?(IIST^^hhrlsl} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((vvb b""$N%N''):*:,,.&/&11346688:t;t==?`@`BBDLELGGI8J8LLN O QQSTV|V|XXZd[d]]_L`Lbbd4e4ggijllnoqxqxssu`v`xxzH{H}}00tt\\ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTS!S!S)%S)%S:S(:S~S~SSS#S(#SSSSSSSS SS SSL SaCS aCSSS SS?\S ?\SSSSSSSSDSSS/lS /lS)S)S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  W@ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  W@ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  W@  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`5Yj}0TCw;Y}_#!EW{?c b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@1U3W5Y7[9];_=a?c bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations13  m J '  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  PP):*:44>%?%IIS T ^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33##o o""$Y%Y'')C*C,,.-/-11346689;v;v==?`@`BBDJEJGGI5J5LLNOQQSTVyVyXXZa[a]]_I`Ibbd1e1ggijllnoququssu]v]xxzE{E}}--uu__ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S˃S˃SuSLuSSTS!S!S %S %S:S(:S~S~SmSmSS(SSSSSeSeSS S}S }SSL SCCS CCS}QS }QSZS ZSSSYSYS=S=S)SD)SSS;lS ;lSeSeS5S05SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 `r" bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 `r" bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 `r"  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`?c:Ko2*Nud,CgBfw5  DUy b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@dJnTx^hr|% bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1f G ( j  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  TT)B*B44>0?0IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffTT B B  00  q q""$_%_'')M*M,,.;/;113)4)6689;;=>@|@|BBDjEjGGIXJXLLNBOBQQS,T,VVXY[[]^`u`ubbd_e_ggiIjIlln3o3qqstvvxy{|{|}}ffۂۄPPŇʼn::$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃS SL SSTS!S!S)%S)%S:S(:S~S~SSS#S(#SSSSSSSS SS SSL SaCS aCSQS QS[S [SOSOSDžSDžSKSKSSDSSS#lS #lS S S߉S0߉S_S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  V> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  V> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  V>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`8nUy(X|Cg{Bf1Ui,PCW{ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@AeCgEiGkImKoMqOs bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 a>x } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  NN)8*844>"?"IIST^y^yhhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''ss] ]""$G%G'')1*1,,./11346z6z88:d;d==?N@NBBD8E8GGI"J"LLNOQ{Q{SSUaVaXXZG[G]]_-`-bbdeggiikglgnnpKqKssu/v/xxz{}}iiۄۆOO‰‹55 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S˃S˃S{SL{SSTS!S!S%S%S:S(:S~S~SsSsS S( SSSSSkSkSS SS SSL SICS ICSRS RS\S \SWSWSȅSȅSSSSDSSS)lS )lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n< b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 as$ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n< b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 as$ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n<  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 as$  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`?c!Ev)4hUy/SDh#Gx6W{ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ /9C)M3W=aGkQ u  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1K , nO } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  QQ);*;44>%?%IIS T ^~^~hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))wwc c""$N%N'')8*8,,."/"113 4 6688:k;k==?U@UBBD?E?GGI)J)LLNOQQSSUhVhXXZN[N]]_4`4bbdeggijlslsnnpYqYssu?v?xxz%{%}} ~~dd׉׋JJ00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSSLSSTS{!S{!S$S$S#:S(#:S~S~S[S[SS(SqSqSSSSSS SkS kSSL S1CS 1CS7RS 7RSs[S s[SSSSSSS)SD)SSS;lS ;lSSSSS#S0#SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ?  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ?  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ?   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`(LqdKo5Yp%IAe|#&Mql2 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@iou{$* 06< bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1>  | [ :  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  QQ)?*?44>/?/IIST^^hissss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccQQ ? ?  --  n n""$\%\'')J*J,,.8/8113&4&6689;;=>@y@yBBDgEgGGIWJWLLNEOEQQS3T3VVX!Y![[]^``bbdtetggibjbllnKoKqqs/t/vvxy{{}}iiۂۄMM11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS ˃S ˃SoSLoSSTS{!S{!S$S$S:S(:S~S~S[S[SS(SqSqSSS_S_SS SkS kSSL S1CS 1CSRS RSU[S U[SSSSSSS7S7S)SD)SSS;lS ;lSGSGSS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a- b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 IU bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a- b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 IU bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a-  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 IU  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` /9]h%#.R}d*NuY}m  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@7[AeKoUy_is} bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1\ =   `  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  ZZ)F*F44>0?0IIST^^histst} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jj^^ R R  FF::**  "#%x%x'')d*d,,.P/P113<4<668(9(;;=>@@BBDpEpGGIXJXLLN@O@QQS(T(VVXY[[]]_l`lbbdTeTggi#?#IIS T ^^hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccMM 7 7  !!  jjT T""$>%>'')(*(,,./11335q6q88:[;[==?E@EBBD/E/GGIJLLNOQzQzSSUdVdXXZN[N]]_8`8bbd"e"ggi j llnnpeqessuIvIxxz-{-}}ggىًKK// b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS˃S˃SuSLuSSTS{!S{!S$S$S:S(:S~S~S[S[SS(SqSqSSSeSeSS SkS kSSL S1CS 1CSgRS gRS[S [S뚅S뚅SȅSȅSSSSDSSS#lS #lS S S߉S0߉SGS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c1 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 HV bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c1 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 HV bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c1  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 HV  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`>sh 2VY} 0?cJn%I$Hb<` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@& 0: D*N4X>bH l  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1A "dE } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  TT)<*<44>"?"IIST^x^xhhr\s\} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhXX H H  88((p p""$X%X'')@*@,,.(/(11346688:l;l==?R@RBBD6E6GGIJLLNNPpQpSSUTVTXXZ8[8]]_`bbdegrgriikVlVnnp:q:ssuvxxz{}t}tXXʄʆ<< b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSSLSSTS!S!S$S$S):S():S~S~SaSaSS(SwSwSSSSSS SqS qSSL S7CS 7CSQS QS [S  [SUSUSSSSS1S1S/SD/SSSAlS AlSkSkS;S0;SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 @  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 @  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 @   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`>lHl2V4>b*\6Z $H*Nw b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@'K3W?cKoW{c o{ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1K -  q S  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  [[)G*G44>3?3IIST^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1gg[[ O O  CC77++  "#%%'')u*u,,.a/a113M4M668999;;=%>%@@BCEEGGIsJsLLN[O[QQSCTCVVX+Y+[[]^``bbdoeoggiWjWlln?o?qqs't'vvxy{{}}kk߂߄SSLJlj;;## b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS˃S˃SiSLiSSTS!S!S$S$S}:S(}:S~S~SaSaSS(SwSwSSSYSYSS SqS qSSL S7CS 7CSQS QS+[S +[SsSsSSSSSASDASSSSlS SlSSSkS0kSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V" b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 L i bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V" b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 L i bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V"  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 L i  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$'K[9nVzzZ~ DDhFj b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@e kqw}$* 06 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1;  y X 7  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  \\)N*N44>@?@IIS,T,^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jj\\ N N  @@22$$  "#%%'')s*s,,.e/e113W4W668I9I;;=;>;@@B-C-EEGHJJLMOOQQStTtVVXbYb[[]P^P``b'?'IIS T ^}^}hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggWW G G  77''  ""$k%k'')S*S,,.;/;113#4#668 9 ;;==?g@gBBDOEOGGI7J7LLNOQQSSUqVqXXZU[U]]_9`9bbdeggijlslsnnpWqWssu;v;xxz{}}uuYYˉˋ==!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS ˃S ˃SoSLoSSTS!S!S$S$S}:S(}:S~S~SaSaSS(SwSwSSSYSYSS SqS qSSL S7CS 7CS+RS +RSg[S g[SSSȅSȅSoSoS)SD)SSS;lS ;lS)S)S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X& b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 K i bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X& b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 K i bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X&  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 K i  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`7Z~:id  DQun9]*Nx! b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@<B$H*N0T6Z<`B f  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 uT3 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  VV)D*D44>2?2IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhVV D D  22  s s""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIZJZLLNDODQQS.T.VVXY[[]^`w`wbbd]e]ggiCjClln'o'qqs t vzvzxxz\{\}}>> ttVV b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃSSLSSTS!S!S%S%S/:S(/:S~S~SgSgSS(S}S}SSS S SS SwS wSSL S=CS =CSQS QS%[S %[SmSmSSSSSSDSSS/lS /lS S S߉S0߉SkS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ? bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ? bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ?  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` 1?c1U#;_#Gx8\9j4X+\ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@`z+!E;_Uyo : bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 o X A *  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  OO)9*944>!?!IIST^z^zhhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((tt^ ^""$H%H'')2*2,,./11346{6{88:e;e==?M@MBBD3E3GGIJLLNNPrQrSSUXVXXXZ>[>]]_$`$bbd e g}g}iikclcnnpIqIssu/v/xxz{}}nnTTƉƋ88 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃScSLcSSTS!S!S%S%Sw:S(w:S~S~SgSgSS(S}S}S梅S梅SSSSSS SwS wSSL S=CS =CSQS QS[S [S[S[SCSCSySySGSDGSSSYlS YlSSS}S0}S%S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 O` bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 O` bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 O`  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H{da,PeGk;__-Q5 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@` r 1C1UCgUyg bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 k P 5   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  ZZ)J*J44>:?:IIS&T&^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjZZ J J  ::**   ""$r%r'')b*b,,.R/R113B4B668292;;=">"@@BCEEGHJzJzLLNfOfQQSRTRVVX>Y>[[]*^*``bceeghjxjxllndodqqsPtPvvxd b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@0TJnd~ /%I? c Y!}!! "s"" bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 |eN7 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  OO)5*544>?IISS]l^lhhrNsN} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **rrX X""$>%>'')$*$,,. / 1}1}335c6c88:I;I==?-@-BBDEGGIIKbLbNNPDQDSSU&V&XXZ[]y]y__a[b[ddf=g=iiklnnpqsrsruuwTxTzz|6}6jjڋڍJJ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS˃S˃SiSLiSSTS!S!S%S%Sw:S(w:S~S~SgSgSS(S}S}S梅S梅SSSSSS SwS wSSL S=CS =CSRS RS[[S [[SSSɅSɅSSS/SD/SSSAlS AlS5S5SS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 N a bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 N a bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 N a  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` 0Hl!nRv+Z~\5Fjf b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@y&8&J8\Jn\ !n!!!" bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 dI. } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  TT)@*@44>,?,IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,zzf f""$R%R'')>*>,,.*/*11346689;x;x==?d@dBBDPEPGGI8?8IIS T ^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1xxhh X X  HH88%%  ""$n%n'')X*X,,.B/B113,4,6689;;=>@u@uBBD_E_GGIHJHLLN0O0QQSTVVXY[t[t]]_\`\bbdDeDggi,j,llnoqqssupvpxxzX{X}}@@((pp b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃS]S ]SSSq:Sq:S~S~SOSOS驄Sp驄SMSMSSSQS`QS[S[SS$ Se S e SMS MS%S %S_lS _lSS SSSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1u} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1u} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1u k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 s Y}?c%I  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  aa)O*O44>;?;IIS%T%^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ssaa O O  ==++ ~ ~""$l%l'')Z*Z,,.H/H113646668$9$;;=>@@BCEwEwGGIcJcLLNMOMQQS7T7VVX!Y![[] ^ ``bbdjejggiTjTlln>o>qqs(t(vvxy{{}}qq[[ЇЉEE// b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS˃S˃SiSiSS$Sw:Sw:S~S~S[S$[S梅S梅SSSSSS$SQSQSZS4ZSST Se S e S5S 5S S  SGlS GlSS S߉S߉S}S} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations12fU b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations12fU b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations12fU b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ?c'K3k  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10vv  aa)K*K44>3?3IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvbb N N  ::&&p p""$Z%Z'')D*D,,../.11346689;w;w==?a@aBBDKEKGGI3J3LLNOQQSSUrVrXXZX[X]]_>`>bbd"e"ggijlxlxnnp\q\ssu@v@xxz${$}}zz^^щыDD** b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃScS cSSSq:Sq:S~S~SOSOS驄Sp驄SMSMSSSQS`QS[S[SS$ Se S e S5S 5S S  SGlS GlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1u} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1u} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1?s  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1u k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 v \Bf*N  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  ``)J*J44>2?2IIST^^hhrqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1tt`` L L  88$$r r""$]%]'')G*G,,.1/111346689;z;z==?d@dBBDNENGGI6J6LLNOQQSTVuVuXXZ[[[]]_A`Abbd'e'ggi j llnnpfqfssuLvLxxz2{2}}qqWWʎʐ== b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃS-S -SSSA:SA:S~S~SgSgSSpSSSSSQS`QSZSZSS$ S_ S _ SGS GSS SYlS YlSqS qSASASS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ae 0eKo  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  bb)P*P44>>?>IIS,T,^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ttbb P P  >>,,  ""$m%m'')[*[,,.I/I113747668%9%;;=>@@BCExExGGIfJfLLNTOTQQSBTBVVX0Y0[[]^``b c eeggiqjqllnZoZqqs>t>vvx"y"{{}~xx\\·Ή@@$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃScS cSSSw:Sw:S~S~S7S7S梅Sp梅SSSSSSS/QS`/QSkZSkZSS$ S_ S _ S;S ;SS SMlS MlSYS YS)S)SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1`} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1`} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1` k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ?c)h.R  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  cc)O*O44>7?7IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ssgg [ [  OOCC33  " # %%'')m*m,,.Y/Y113E4E668191;;=>@@BCEwEwGGI_J_LLNGOGQQS/T/VVXY[[]]_s`sbbd[e[ggiCjClln+o+qqstvvxxzo{o}}WW˂˄??$$ }} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃS-S -SSS;:S;:S~S~SgSgSSpSSSSSQS`QSZSZSS$ S_ S _ S5S 5S S  SGlS GlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ?c,P<(z  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10rr  \\)F*F44>0?0IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1rr\\ F F  00yyc c""$M%M'')7*7,,.!/!113 4 6688:j;j==?T@TBBD>E>GGI(J(LLNOQQSSUqVqXXZ[[[]]_E`Ebbd/e/ggijllnoqrqrssuVvVxxz:{:}}ttXXʎʐ<< b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃScS cSSSw:Sw:S~S~S7S7S梅Sp梅SSSSSSS/QS`/QSkZSkZSS$ S_ S _ S#S #SS S5lS 5lS S  S߉S߉SYS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1a} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1a} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1a k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ?c)h.R  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  __)G*G44>+?+IIST^^hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1sscc S S  CC33 { {""$c%c'')K*K,,.3/311346689;w;w==?[@[BBD?E?GGI#J#LLNOQyQySSU]V]XXZA[A]]_%`%bbd e g{g{iik_l_nnpCqCssu'v'xxz { }}}}aaӄӆCC%% b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS'S 'SSS;:S;:S~S~S[S[SSpSSSSSQS`QSZSZSS$ SY S Y SAS ASS SSlS SlSwS wSGSGSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Dh8}cIm  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  hh)T*T44>>?>IIS&T&^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1tthh \ \  PPDD88,,  " # %%' ( **,,.n/n113Z4Z668F9F;;=2>2@@BCEEG H J~J~LLNfOfQQSNTNVVX6Y6[[]^``bcezezggibjbllnJoJqqs2t2vvxy{{}~vv^^҇҉FF.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS˃S˃SiS iSSS}:S}:S~S~SOSOSSpSYSYSSS)QS`)QSeZSeZSS$ SY S Y SGS GSS SYlS YlSS SqSqSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1}} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1}} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1} k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ?c)h.R  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10oo  aa)S*S44>C?CIIS/T/^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ooaa S S  EE77))  " # %%'')x*x,,.j/j113\4\668N9N;;=@>@@@B2C2EEG$H$JJLMOOQRTwTwVVXeYe[[]S^S``b?c?eeg)h)jjlmooqqsltlvvxRyR{{}8~8ttXXʑʓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS!S !SSS5:S5:S~S~SaSaSSpSSSSSQS`QSZSZSS$ SY S Y S)S )SS S;lS ;lS S  S߉S߉SwS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ?c'K3k  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10rr  bb)J*J44>0?0IIST^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1rrbb R R  BB22""  "#%v%v'')^*^,,.F/F113.4.6689;;==?r@rBBDZEZGGI@J@LLN$O$QQSTVzVzXXZ^[^]]_B`Bbbd&e&ggi j l|l|nnp`q`ssuDvDxxz({(}} ~~bbԉԋFF** b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS˃S˃SiS iSSS}:S}:S~S~SOSOSSpSYSYSSSQS`QSZSZSS$ SY S Y S/S /SS SAlS AlS/S /S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1|} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1|} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1| k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ?c)M7!r  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10oo  ]])K*K44>7?7IIST^^hisqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1oo]] K K  99'' z z""$h%h'')V*V,,.D/D113242668 9 ;;=>@@BBDsEsGGI_J_LLNIOIQQS3T3VVXY[[]^`|`|bbdbebggiHjHlln,o,qqstvvxxza{a}}CC%%wwYY b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS!S !SSS5:S5:S~S~SySySSpSSSSSQS`QSZSZSS$ SY S Y SS SS S/lS /lS S  S߉S߉SkS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1&} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1&} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1& k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 qOs-Q /z  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  XX)B*B44>(?(IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mmYY E E  11}}g g""$Q%Q'');*;,,.%/%11346688:n;n==?T@TBBD:E:GGI J LLNOQyQySSU_V_XXZE[E]]_+`+bbdeggiikjljnnpPqPssu6v6xxz{}}uuYYˉˋ==!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS ˃S ˃SoS oSSS}:S}:S~S~SSSSpSYSYSSS/QS`/QSkZSkZSS$ SY S Y SAS ASS SSlS SlSS SwSwSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z& b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Q} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z& b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Q} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z& b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Q k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 @d,l4X!  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  [[)K*K44>9?9IIS%T%^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kk[[ K K  ;;++   ""$s%s'')c*c,,.S/S113C4C668393;;=#>#@@BCEEGHJyJyLLNeOeQQSQTQVVX=Y=[[])^)``bceeghjwjwllncocqqsOtOvvx;y;{{}'~'uuaaבד b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃSS SSS/:S/:S~S~SySySSpS S SSSeQS`eQSZSZSS$ SY S Y S S  SS SlS lS S  S]߉S]߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1'} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1'} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1' k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 f2V"d0T  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  VV)<*<44>?IIST^q^qhhrQsQ} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mmYY E E  11yy_ _""$E%E'')+*+,,./11335j6j88:P;P==?2@2BBDEGGIIKgLgNNPIQISSU+V+XXZ [ ]~]~__a`b`ddfBgBiik$l$nnpqswswuuwYxYzz|;};mm݋ݍMM b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS ˃S ˃SoS oSSS:S:S~S~SSSSpS_S_SSSQS`QS [S [SS$ SY S Y S)S )SS S;lS ;lS5S 5SSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z( b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Q} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z( b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Q} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z( b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Q k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ?c5Y+O!E  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  WW)C*C44>-?-IIST^^hhrqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kkWW C C  //}}i i""$U%U'')A*A,,.-/-11346689;{;{==?g@gBBDSESGGI=J=LLN%O%QQS T VVXXZi[i]]_Q`Qbbd9e9ggi!j!lln o q}q}ssuevexxzM{M}}55yyaa b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃSS SSS):S):S~S~S[S[SSpSSSSSQS`QSZSZSS$ SS S S S)S )SS S;lS ;lS/S /S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 l@d8~ Rv  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  YY)C*C44>,?,IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ll\\ L L  <<,, x x""$b%b'')L*L,,.6/6113 4 668 9 ;;==?i@iBBDSESGGI/?/IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggUU C C  11  r r""$`%`'')N*N,,.@}@}BBDkEkGGIWJWLLNAOAQQS+T+VVXY[[]]_t`tbbd^e^ggiHjHlln2o2qqstvvxy{{{{}}eeڂڄOOćĉ99## b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃSS SSS):S):S~S~S[S[SSpSSSSSSQS`SQSZSZSS$ SS S S SS SS S)lS )lS S  S߉S߉SeS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 n4X _%I  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  UU)?*?44>'?'IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjVV B B  ..zzd d""$N%N'')8*8,,."/"113 4 6688:k;k==?U@UBBD?E?GGI'J'LLN O QQSSUfVfXXZL[L]]_2`2bbdeggiiklllnnpPqPssu4v4xxz{}}nnRRʼnŋ88 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS˃S˃SuS uSSS:S:S~S~SCSCSSpSeSeSSSQS`QSZSZSS$ SS S S SS SS S/lS /lS S  S߉S߉SwS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c1 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1y} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c1 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1y} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c1 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1y k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ?c1U#G9  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  TT)>*>44>&?&IIS T ^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,zzf f""$Q%Q'');*;,,.%/%11346688:n;n==?X@XBBDBEBGGI*J*LLNOQQSSUiViXXZO[O]]_5`5bbdeggijltltnnpZqZssu@v@xxz&{&}} ee؉؋KK11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃSS SSS#:S#:S~S~SmSmSSpSSSSSAQS`AQS}ZS}ZSS$ SM S M S)S )SS S;lS ;lSSS SS#S#SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1#} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1#} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1# k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 c)MRv<  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  VV)D*D44>2?2IIS T ^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhVV D D  22  s s""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIZJZLLNHOHQQS6T6VVX$Y$[[]^``bcewewggiejellnNoNqqs2t2vvxy{{}}llނބPP‡‰44 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S˃S˃SuS uSSS:S:S~S~SSSSpSeSeSSSORS`ORS[S[SS$ SM S M SS SS S/lS /lS;S ;S S SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Q} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Q} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Q k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  .6>#G  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  WW)C*C44>+?+IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1gg[[ O O  CC77''  ""$u%u'')a*a,,.M/M113949668%9%;;=>@@BBDkEkGGISJSLLN;O;QQS#T#VVX Y [[]]_g`gbbdOeOggi7j7llnoqqstv{v{xxzc{c}}KK33qq b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃSS SSS#:S#:S~S~SmSmSSpSSSSSQS`QS[S[SS$ SM S M SS SS S)lS )lS S  S߉S߉SqS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1$} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1$} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1$ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 y_Ei+O  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  PP):*:44>$?$IIST^^hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffPP : :  $$mmW W""$A%A'')+*+,,./11335t6t88:^;^==?H@HBBD2E2GGIJLLNOQ{Q{SSUeVeXXZO[O]]_9`9bbd#e#ggi j llnnpfqfssuJvJxxz.{.}}hhډڋLL00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S˃S˃S{S {SSS:S:S~S~SSSSpSeSeSSSQS`QS[S[SS$ SM S M S S  SS SlS lS S  S߉S߉S;S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l: b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Q} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l: b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Q} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l: b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Q k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 s Y}?c%I  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  SS);*;44>?IIST^u^uhhrWsW} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggWW G G  77''o o""$W%W'')?*?,,.'/'11346688:k;k==?O@OBBD3E3GGIJLLNNPmQmSSUQVQXXZ5[5]]_`bbddfogoiikSlSnnp7q7ssuvxxzz|q}qUUDŽdž77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYSʃSʃS S  SSS:S:S~S~SOSOSSpSSSSSkQS`kQSZSZSS$ SM S M S)S )SS S;lS ;lS_S _S/S/SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 j:^ .rBf  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  \\)H*H44>2?2IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hh\\ P P  DD88,,   "#%%'(*v*v,,.b/b113N4N668:9:;;=&>&@@BCEEGGIrJrLLNZOZQQSBTBVVX*Y*[[]^``bbdnenggiVjVlln@o@qqs(t(vvxy{{}}llTTȇȉ<<$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYS˃S˃S{S {SSS:S:S~S~S=S=SSpSkSkSSSQS`QSZSZSS SM S| M S5S 5S]~S ]~S;NS ;NSS SމSމS-S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uA  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1w} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uA  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1w} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uA  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1w k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1"} {y  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  UU)G*G44>7?7IIS#T#^^hiss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccUU G G  99++  "#%z%z'')l*l,,.^/^113P4P668B9B;;=4>4@@B&C&EEGHJJLMO}O}QQSkTkVVXYYY[[]G^G``b3c3eeghjjlmooqqsjtjvvxPyP{{}6~6rrVVȑȓ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYSʃSʃS S  SSS:S:S{~S{~SOSOSSpSSSSSgRS`gRS[S[SS$ SM S M SS SS S#lS #lS S  S߉S߉S_S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  /A/SAe  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  VV)>*>44>$?$IIST^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&& ~ ~""$j%j'')R*R,,.:/:113"4"668 9 ;~;~==?f@fBBDNENGGI4J4LLNOQQSSUnVnXXZR[R]]_6`6bbdeggiikplpnnpXqXssu>"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYS˃S˃SS SSS:S:S~S~S=S=SSpSkSkSSSkQS`kQSZSZSS SM S| M SS SE~S E~S#NS #NSS SމSމSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uC b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1v} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uC b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1v} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uC b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1v k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1z`FjCg  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  QQ)?*?44>+?+IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccQQ ? ?  --  n n""$\%\'')J*J,,.8/8113&4&6689;;=>@y@yBBDgEgGGISJSLLN=O=QQS'T'VVXY[[]]_p`pbbdVeVggi?IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%ss] ]""$G%G'')1*1,,./11346z6z88:d;d==?J@JBBD0E0GGIJLLNNPoQoSSUUVUXXZ;[;]]_!`!bbdeggiiklllnnpRqRssu8v8xxz{}}ww[[͉͋??## b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYS˃S˃SS SSS:S:S~S~SSS Sp SqSqSSSkQS`kQSZSZSS SG S| G S;S ;Sc~S c~SANS ANSS SމSމS?S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~J b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Q} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~J b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Q} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~J b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Q k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1xhX|Hl  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  OO)I*I44>7?7IIS#T#^^hiss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__OO ? ?  //w w""$q%q'')a*a,,.Q/Q113A4A668191;;=!>!@@BCEEGHJwJwLLNcOcQQSOTOVVX;Y;[[]'^'``bceeggiujullnmomqqsYtYvvxEyE{{}1~1 kk b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYSʃSʃSS SSS:S:Su~Su~SaSaSSpSSSSSQS`QSZSZSS SG S| G SS S'~S '~SNS NS S  Sw݉Sw݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1!} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1!} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1@t b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1! k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1Z~VzRvl  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  NN)4*444>?IIST^u^uhhrUsU} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%qqW W""$=%='')#*#,,. / 1|1|335b6b88:H;H==?*@*BBD E G}G}IIK_L_NNPAQASSU#V#XXZ[]v]v__adbdddfFgFiik(l(nnp q s{s{uuw]x]zz|?}?!!qqQQ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYS˃S˃SS SSS:S:S~S~SSS Sp SqSqSSSQS`QSSZSSZSS SG S| G S#S #SK~S K~S)NS )NS)S )SމSމSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~L b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Q} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~L b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Q} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~L b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Q k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1'f,P  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  KK)?*?44>)?)IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__KK 7 7  ##qq] ]""$Q%Q'')=*=,,.)/)11346689;w;w==?c@cBBDOEOGGI9J9LLN!O!QQS T V}V}XXZe[e]]_M`Mbbd5e5ggijllnoqqssumvmxxzU{U}}==%% ii b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations1MSoʃSoʃSS|SySySSS"S|"S &SH &S9S9SK~SPK~SSXSoSoSSPSSS_S_SS$ SiS, iSS SS$ SS SEDS| EDSRSXRS[S|[S S SƅSxƅSS\S?S?S='S='S#S8#SSS5lS5lSySySASlASSSS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Pu b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations11 bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Pu b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations11  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Pu  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations11 bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`AZ~n$RvAeJn$HeBfHl?l b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@7[X|Pt=aHlCg@dIm8\Os0TUy(L[ Da< 4   ,   $ |   t   l  d\X|TxPtLpHlDh@d<`8\4X0T,P(L$H D@ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d7X0u`T83h@pJ  $ ~ X 2   f@t bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10==  QQ)G*G44>1?1IIS%T%^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1==11 % %  ==11!!   ""$y%y'')e*e,,.Q/Q113=4=668)9);;=>@@BBDqEqGGIYJYLLNAOAQQS)T)VVXY[[]]_m`mbbdaeaggiIjIlln1o1qqstvvxy{u{u}}]]тфEE,,# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xMS5˃S5˃SSdSSlSc!Sc!S$S$S:S@:SSSCSCSېS@ېSYS YS%S%SSSS SSS SSSd SCS CS%RS %RSa[S a[SSS̅S̅S]S]S S S )S  )SSSSpS#lSt#lSySTyS S S߉S$߉SYS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h4 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 A bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h4 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 A bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h4  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 A  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`c8\ 8m\8w\ 8\* b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@9]o&3W(L-QNr'K!Et+?-Q93Sw y 0  2 V  X |   ~  5  7[]:<`f%IOsy28\b!EKou .4 X  ^!!!"""#A###G$k$$% b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d9o^kMD<+ xgV  G  8  )    |m^O@ bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  NN)*44> ? IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%ss] ]""$G%G'')1*1,,./1l1l335V6V88:@;@==?L@LBBD2E2GGIJLLNNPqQqSSUWVWXXZ=[=]]_#`#bbdeggiiknlnnnpTqTssu:v:xxz { }}yy__щыCC''  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1MSoʃSoʃSS|SySySSS"S|"S &SH &S9S9SE~SPE~SSXSoSoSSPSSSYSYSS$ SiS, iSS SS$ SS SEDS| EDScSSXcSS\S|\S盅S盅S҅Sx҅S]S\]SSS*S*S S8 SSSlSlSySyS Sl S߉S߉SAS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Or b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations11 bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Or b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations11  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Or  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations11 bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`3Wz6Z<o 1Tx=aIm .REi b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@6Z=aDhKog R v m  #   s  Y } y  *    & `  1,2g88n?u"F|'K}(L~)M*N+O, P  !!-"Q"""##.$R$$$%%/&S&&&''0(T((())1*U**+ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d $-6qP?/HQ  Z  a  h  o  v}   bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10<<  MM)=*=44>#?#IIST^^hhrisi} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1<<,,    11!!  i i""$Y%Y'')A*A,,.)/)11346688:m;m==?S@SBBD7E7GGIJLLNNPqQqSSUUVUXXZ9[9]]_`bbd e ggiikclcnnpGqGssu+v+xxz{}}eeׄ׆II++ # b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xMS;˃S;˃SSdSSlSc!Sc!S$S$S:S@:SSSCSCSېS@ېSYS YS%S%SSSS SSS SSSd SCS CSSS SS\S \SSSЅSЅSSS]S]S[*S [*SSSSpSlStlSySTyS S SQ߉S$Q߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 A bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 A bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 A  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`4XDh$H| Dm`5i8 1Y}!| b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@(Lt>f 0X|"Jn /  )  < ` # |     v . R  p   z j  D dl^6^(Pt@d,Pt<`(Lp8\$ H  l!!"4"""X#|## $$$D%h%% &&&0'T'''x(()@) b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d v O(Z           bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  NN)*44>?IIST^w^whhrYsY} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%qqW W""$=%='')#*#,,. / 1W1W335=6=88:#;#==?,@,BBDEGGIIKaLaNNPCQCSSU%V%XXZ[]x]x__afbfddfHgHiik*l*nnp q s}s}uuw_x_zz|A}A##uuUUŐŒ  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSʃSʃSSdSSlSi!Si!S$S$S:S@:S{~S{~SISISᐄS@ᐄS_S _SSSSSS SYS YSSd SCS CSRS RS[S [S9S9SąSąSeSeSSS&S &S/S/SSpSAlStAlSySTySSSYS$YSS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ,A bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ,A bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ,A  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` DAet 0CPt  -<`~<hl(Hl b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@4XOs5Y6Z?8\%:^t <`Z~@d>b&J@d 0BfeK o  1 U   ;   ! p   V z  <`&J4oY}Cg-Q;%v`Jn4XB,}g b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d4Omn<K (Z)xGg8   Z  + | M o@ bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  SS)E*E44>?IIS#T#^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaSS E E  77))   ""$x%x'')j*j,,.\/\113N4N668@9@;;=2>2@@B$C$EEGHJkJkLLNYOYQQSGTGVVXYYY[[]G^G``b3c3eeghjjlmooqqsjtjvvxPyP{{}6~6ttXXʑʓ  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSS/˃S/˃SSdSSlSi!Si!S$S$S:S@:SSSISISᐄS@ᐄS_S _SSS}S}SS SYS YSSd SCS CSIRS IRS[S [S͚S͚SgʅSgʅS#S#SYSYSW(S W(SSSSpS)lSt)lSySTyS#S#S߉S$߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]) b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 HS  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]) b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 HS  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1])  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 HS   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Bf=f2-Qtz(@dAe 1<Uy,P b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@4Xf2Vz(0T<.R,P,P@d*NTx(L&Jh| *   >  . R  B f  V z  j~,@4XLpd|, D8\Pth 0$H<`Tx l  !!!"4""" b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d4fLj2E ~dJ0|d L 4    lT<$  bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  QQ);*;44>&?&IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$p p""$Z%Z'')D*D,,../.11346689;w;w==?a@aBBDKEKGGI6J6LLNOQQSTVzVzXXZb[b]]_J`Jbbd4e4ggijllnoqxqxssu`v`xxz${$}} xxbb  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSʃSʃSSdSSlSi!Si!S$S$S:S@:Su~Su~SISISᐄS@ᐄS_S _SSSSSS SYS YSSd SCS CSRS RS\S \S]S]SgąSgąS#S#SySySw&S w&SSSSpS)lSt)lSySTySSS߉S$߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 +A bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 +A bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 +A  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`2Vhx8|j**Lp> >b.R Dv b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Pt8\ D,|dLp4X@(Dhx ` F j  H l H l  0 T J n   < L p   & N r v   ^PtBfRv&J .|`Dh(L 0~bFj*N2dHl,P b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dP p@`0bAP {n<   X  & t B ^, bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  OO)=*=44>?IIST^^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaOO = =  ++~~l l""$Z%Z'')H*H,,.6/6113$4$6689;;=>@w@wBBDeEeGGI.J.LLNOQQSTVVXY[[]]_p`pbbdVeVggi |  <Rv(L:"i?c9Vz,P&mCg=Z~0T*q b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dI[$m6{VH1 Zg!h  / v =   K  Y  bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  MM)7*744>!?!IIS T ^|^|hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&rr\ \""$F%F'')0*0,,./11346y6y88:c;c==?M@MBBD7E7GGI!J!LLNOQzQzSSU`V`XXZF[F]]_,`,bbdeggiikjljnnpNqNssu2v2xxzz|c}cՁGGRRʼnŋ88  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃS SL SSTS{!S{!S$S$S:S(:S{~S{~S[S[SS(SqSqSSSSSS SkS kSSL S1CS 1CSRS RS'\S '\SoSoSaSaSaSaS)SD)SSS;lS ;lSSS_S0_SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 A bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 A bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 A  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ae#Gz<}_?!TxW{9].Ru b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@'K%I#G!ECA?= bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1']8 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  QQ)A*A44>1?1IIST^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaQQ A A  11!! y y""$i%i'')Y*Y,,.I/I113949668)9);;=>@@B C EEGGIqJqLLN]O]QQSITIVVX5Y5[[]!^!``b c eeggiojolln[o[qqsGtGvvx3y3{{}~ mmYYϑϓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS)˃S)˃SSLSSTS!S!S$S$S:S(:S~S~SaSaSS(SwSwSSSwSwSS SqS qSSL S7CS 7CSRS RS[S [S!S!SSSmSmSSDSSS/lS /lSMSMSS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Sw bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Sw bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Sw  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`AeBp(7[Fj.R,Pn$H&Tx *N| b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@+O-Q/S1U3W5Y7[9] bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1+eB } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  RR)@*@44>0?0IIST^^histst} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddRR @ @  ..  o o""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGIXJXLLNFOFQQS4T4VVX"Y"[[]^``bbdueuggicjcllnLoLqqs0t0vvxy{{}}jj܂܄NN22 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLSSTS{!S{!S$S$S:S(:S{~S{~S[S[SS(SqSqSSSSSS SkS kSSL S1CS 1CSRS RS \S  \SQSQSSSSSSSSDSSS)lS )lSSS߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 A bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 A bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 A  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`1UAev24e#Dhm3WW{x6g b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@; 9  7  5  3 1 / - bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1nI$Z } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  MM)9*944>%?%IIS T ^^hhrisi} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%ss_ _""$K%K'')7*7,,.#/#11346688:q;q==?]@]BBDIEIGGI5J5LLNOQQSTVyVyXXZa[a]]_I`Ibbd1e1ggijllnoququssu]v]xxzE{E}}--qqYY b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS)˃S)˃SSLSSTS!S!S$S$S:S(:SSSaSaSS(SwSwSSS}S}SS SqS qSSL S7CS 7CS]SS ]SS\S \SᛅSᛅSSSSS SD SSSlS lS S S߉S0߉SkS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V" b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Tx bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V" b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Tx bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V"  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Tx  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Hl:^  D4X>bd<`8\Bf<` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@;_= a ? c C g E i G k ImKo bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1|Y6p } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  LL)6*644>"?"IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbLL 6 6      iiS S""$=%='')'*',,./11335p6p88:Z;Z==?D@DBBD.E.GGIJLLNOQyQySSUcVcXXZM[M]]_7`7bbd!e!ggi j llnnpdqdssuHvHxxz,{,}}ff؉؋JJ.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLSSTS!S!S %S %S:S(:S~S~SmSmSS(SSSSSSSS S}S }SSL SCCS CCS!SS !SS]\S ]\SSS=S=SOSOS#SD#SSS5lS 5lSYSYS)S0)SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 !C  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 !C  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 !C   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ae&J$Hin!6l6%I\$MqZ~" b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@>82,& ysm bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1sL%~ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  OO)=*=44>+?+IIST^^hhrtst} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaOO = =  ++~~l l""$Z%Z'')H*H,,.6/6113$4$6689;;=>@w@wBBDeEeGGISJSLLN=O=QQS'T'VVXY[[]]_p`pbbdZeZggiDjDlln.o.qqstvvxy{w{w}}aaւքKK55 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS˃S˃SSLSSTS!S!S%S%S:S(:S~S~SySySS(SSS S SqSqSS SS SSL SOCS OCSRS RS-\S -\SuSuSSSmSmS)SD)SSS;lS ;lS_S_S/S0/SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yG b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ^O. bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yG b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ^O. bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yG  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ^O.  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ae#Gz<(L_2!Tx9]g.Ru b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@"F(L.R4X:^@dFjLp bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1"`? } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  \\)H*H44>4?4IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hh\\ P P  DD88,,   "#%%'(*v*v,,.b/b113N4N668:9:;;=&>&@@BCEEGGItJtLLN\O\QQSDTDVVX,Y,[[]^``bbdpepggiXjXlln@o@qqs(t(vvxy{{}}llTTȇȉ<<$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃS SL SSTS!S!S%S%S:S(:S~S~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CSRS RS \S  \SQSQS S SSSSDSSS#lS #lS S S߉S0߉SkS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  B bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  B bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  B  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ae0Tt@ 1a*Ae{/Seu3c b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@g a  [  U y O s I m Cg=a bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1W0 b; } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  NN)8*844>"?"IIST^{^{hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&tt` `""$K%K'')5*5,,./113 4 6~6~88:h;h==?R@RBBDbEi,BfOs!Lp b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@RvX | ^  d j  p v|# bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1}\;y } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  VV)>*>44>&?&IIS T ^|^|hhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&& ~ ~""$j%j'')R*R,,.:/:113"4"668 9 ;~;~==?f@fBBDNENGGI6J6LLNOQQSSUpVpXXZT[T]]_8`8bbdeggijlrlrnnpVqVssu:v:xxz{}}ttXXʉʋ<<  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTS!S!S/%S/%S#:S(#:S~S~SSS)S()SSSSSSS S  SS SSL SgCS gCSQSS QSS\S \S՛S՛SSS=S=S#SD#SSS5lS 5lS;S;S S0 SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  Y@ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  Y@ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  Y@  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`)Me/S0T}'K^(L.R{%I,P b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ 1'~tj `VzLpBf bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 d;i } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  UU)A*A44>+?+IIST^^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeYY M M  AA55%%  ""$s%s'')_*_,,.K/K113747668#9#;;=>@@BBDkEkGGISJSLLN;O;QQS#T#VVX Y [[]]_g`gbbdOeOggi7j7llnoqqstv{v{xxzc{c}}KK33ss b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S˃S˃S{SL{SSTS!S!S%S%S:S(:S~S~SsSsS S( SSSSSkSkSS SS SSL SICS ICSSS SS?\S ?\SSSSSsSsSSDSSS)lS )lS S S߉S0߉S_S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n< b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ^l" bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n< b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ^l" bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n<  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ^l"  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$Hw|@ad(Ko\!5YEi =a b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@=#G-Q7[AeKoUy_ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1z[< } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  PP):*:44>"?"IIST^{^{hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))uu_ _""$I%I'')3*3,,./11346|6|88:f;f==?N@NBBD4E4GGIJLLNOQsQsSSUYVYXXZ?[?]]_%`%bbd e g~g~iikdldnnpJqJssu0v0xxz{}}ooUUljNj99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTS!S!S)%S)%S#:S(#:S~S~SSS#S(#SSSSSSSS SS SSL SaCS aCSRS RS[S [S9S9SöSöSoSoS SD SSSlS lS S S߉S0߉SAS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  X> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  X> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  X>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`/Syr,y[4XyAy`Im b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@8\. R $ H  >  4  * w bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1@nE } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  QQ)9*944>?IIST^u^uhhrYsY} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%%m m""$U%U'')=*=,,.%/%113 4 6688:i;i==?O@OBBD3E3GGIJLLNNPmQmSSUQVQXXZ5[5]]_`bbddfogoiikSlSnnp7q7ssuvxxzz|q}qUUDŽdž99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S˃S˃SSLSSTS!S!S%S%S:S(:S~S~SsSsS S( SSS S SqSqSS SS SSL SICS ICSSS SS?\S ?\SSSSSSSSDSSSlS lS S SQ߉S0Q߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p@  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 _m$ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p@  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 _m$ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p@   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 _m$  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`4X~| @zt8v\ tTx<` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@i s  }  & 0  : D*N4X bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1~_@" } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  NN)4*444>?IISS]k^khhrMsM} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))qqW W""$=%='')#*#,,. / 1|1|335b6b88:H;H==?,@,BBDEGGIIKaLaNNPCQCSSU%V%XXZ[]x]x__aZbZddf;?;IIS'T'^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeWW I I  ;;--  "#%|%|'')n*n,,.`/`113R4R668D9D;;=6>6@@B(C(EEGHJJL M OOQQSoToVVX]Y][[]K^K``b7c7eeg!h!jjl m o~o~qqsdtdvvxJyJ{{}0~0nnRRđē b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS˃S˃SuSLuSSTS{!S{!S$S$S:S(:S~S~S[S[SS(SqSqSSSeSeSS SkS kSSL S1CS 1CSRS RS[S [SSSSSmSmS)SD)SSS;lS ;lS5S5SS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c1 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 GS bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c1 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 GS bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c1  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 GS  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`An2$Ae9$He=HlA?l b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@4>$H.R8\BfLpVz bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1qR3 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  UU)?*?44>*?*IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhXX H H  88((t t""$^%^'')H*H,,.2/211346689;{;{==?e@eBBDOEOGGI:J:LLN"O"QQS T V~V~XXZf[f]]_N`Nbbd6e6ggijllnoqzqzssubvbxxzJ{J}}22zzdd b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSSLSSTS{!S{!S$S$S):S():S~S~S[S[SS(SqSqSSSSSS SkS kSSL S1CS 1CSRS RS[S [S S SSSWSWS#SD#SSS5lS 5lS#S#S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 A  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 A  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 A   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`,Pel /SLp@_l )M<`@\ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ /! v  h  Z ~ L p > b 0T bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1)~S(} } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>/?/IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIWJWLLNAOAQQS+T+VVXY[[]]_t`tbbdZeZggi@j@lln$o$qqstvwvwxxzY{Y}};;qqSS b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS˃S˃S{SL{SSTS{!S{!S$S$S:S(:S~S~S[S[SS(SqSqSSSeSeSS SkS kSSL S1CS 1CSURS URS[S [SٚSٚSSSSSSDSSS)lS )lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e5 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 FT bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e5 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 FT bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e5  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 FT  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`)Mf=akz(UyQu?c<e,P b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@` j  t  ~  ' 1 ;!E+O bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1uV7y } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  QQ);*;44>%?%IIS T ^|^|hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **vv` `""$J%J'')4*4,,./11346}6}88:g;g==?Q@QBBD;E;GGI%J%LLN O Q~Q~SSUdVdXXZJ[J]]_0`0bbdeggiikjljnnpNqNssu2v2xxz{}}llބކRRʼnŋ88 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSSLSSTS!S!S$S$S/:S(/:S~S~SaSaSS(SwSwSSS S SS SqS qSSL S7CS 7CS[RS [RS[S [SߚSߚSSS[S[SASDASSSSlS SlSSSwS0wSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 B  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 B  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 B   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`=mf7[-Qz%(fAe0T<Uy,P b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$,4< D(L0T8\ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1`@  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  UU)E*E44>5?5IIS!T!^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%% } }""$m%m'')]*],,.M/M113=4=668-9-;;=>@@B C EEGGIuJuLLNaOaQQSMTMVVX9Y9[[]%^%``bceeggisjslln_o_qqsKtKvvx7y7{{}#~#qq]]ӑӓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS ˃S ˃SoSLoSSTS!S!S$S$S:S(:S~S~SaSaSS(SwSwSSS_S_SS SqS qSSL S7CS 7CS+RS +RSg[S g[SSSSS[S[S5SD5S S SGlS GlS_S_S/S0/SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X& b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 J c bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X& b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 J c bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X&  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 J c  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`:`!Buh&Jhp.RZ~x  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@' -39?!E'K-Q bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1bA  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  VV)D*D44>4?4IIS"T"^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhVV D D  22  s s""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGI\J\LLNJOJQQS8T8VVX&Y&[[]^``bceyeyggigjgllnPoPqqs4t4vvxy{{}}nnRRćĉ66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSSLSSTS!S!S%S%S):S():S~S~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CSQS QS[S [SISIS׷S׷SSS)SD)SSS;lS ;lS/S/S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 B bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 B bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 B  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` DhSw22VAe  a1Ut-QBb b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@dH l P t X | `  h px  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1`@ ` } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  QQ)=*=44>)?)IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))wwc c""$O%O'');*;,,.'/'11346688:u;u==?a@aBBDMEMGGI9J9LLN!O!QQS T V}V}XXZe[e]]_M`Mbbd5e5ggijllnoqyqyssuavaxxzI{I}}11uu]] b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS˃S˃SuSLuSSTS!S!S$S$S:S(:S~S~SaSaSS(SwSwSSS_S_SS SqS qSSL S7CS 7CSQS QS1[S 1[SySySSSSS#SD#SSS5lS 5lSSS߉S0߉S}S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z* b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 I c bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z* b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 I c bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z*  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 I c  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`BPt8pFj  .\<`$Hl2V{ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@3W9 ] ? c E i K o Q u W{] bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1^={Z } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  PP):*:44>&?&IIST^^hhrfsf} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffPP : :  $$mmW W""$A%A'')+*+,,./11335t6t88:^;^==?H@HBBD2E2GGIJLLNOQ}Q}SSUgVgXXZQ[Q]]_;`;bbd%e%ggijllnnphqhssuLvLxxz0{0}}jj܉܋NN22 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS!SL!SSTS!S!S%S%S5:S(5:S~S~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CS+RS +RSg[S g[SSS?S?SSS5SD5S S SGlS GlSqSqSAS0ASS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  A bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  A bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  A  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`4fVz1U3` =`j+O#Gt b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@%C=a[y,&JDhb bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1lWB- } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>/?/IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIWJWLLNAOAQQS+T+VVXY[[]]_t`tbbd^e^ggiHjHlln2o2qqstvvxy{{{{}}eeڂڄOOćĉ99## b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS˃S˃SiSLiSSTS!S!S%S%Sw:S(w:S~S~SgSgSS(S}S}S梅S梅SSSSSS SwS wSSL S=CS =CSRS RSU[S U[SSSSSSS;SD;SSSMlS MlSqSqSAS0ASS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 M^ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 M^ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 M^  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`3Rv~,hX|2Rv^<`8d  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$:,PBfX|n3%I bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1gN5 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  ``)L*L44>8?8IIS T ^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ll`` T T  HH<<00$$  "#%%'(*z*z,,.f/f113R4R668>9>;;=*>*@@BCEEGHJxJxLLN`O`QQSHTHVVX0Y0[[]^``bcetetggi\j\llnDoDqqs,t,vvxy{{}}ppXẊ̉@@(( b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS!SL!SSTS!S!S %S %S/:S(/:S~S~SmSmSS(SSSSS S SS S}S }SSL SCCS CCSQS QSZS ZS%S%S-S-S#S#S#SD#SSS5lS 5lSSS߉S0߉S}S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  A bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  A bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  A  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`6i8\ 3W"Fq!  0b[,PFjl b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@  3 - Q K o i :4XRv bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1nYD } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  RR)<*<44>&?&IIS T ^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **xxd d""$O%O'')9*9,,.#/#113 4 6688:l;l==?V@VBBD@E@GGI*J*LLNOQQSSUiViXXZO[O]]_5`5bbdeggijltltnnpZqZssu@v@xxz&{&}} ee؉؋KK11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS˃S˃SiSLiSSTS!S!S%S%S}:S(}:S~S~SgSgSS(S}S}SSSYSYSS SwS wSSL S=CS =CSRS RSC[S C[SSSSSSS#SD#SSS5lS 5lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 L _ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 L _ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 L _  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`7:^!),Pu~Bep 4Uyb b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@; _ Q u g  }  , B4XJn bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1jQ8 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  ZZ)B*B44>*?*IIST^^hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjZZ J J  ::**   ""$n%n'')V*V,,.>/>113&4&6689;;==?j@jBBDRERGGI:J:LLNOQQSTVtVtXXZX[X]]_<`v>xxz"{"}}xx\\Ή΋@@$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃS]S]SS$Sq:Sq:S~S~S[S$[S驄S驄SMSMSS$SQSQSZS4ZSST Se S e SGS GSS SYlS YlSeS eS5S5SS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1-aY b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1-aY b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1-aY b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ae&J4o  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10zz  jj)V*V44>>?>IIS&T&^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1zznn b b  VVJJ::&&  "#%%'')t*t,,.`/`113L4L668898;;=">"@@B C E~E~GGIfJfLLNNONQQS6T6VVXY[[]^`z`zbbdbebggiJjJlln2o2qqstvvxy{v{v}}^^҂҄FF++ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃScS cSSSw:Sw:S~S~SOSOS梅Sp梅SSSSSSSGQS`GQSZSZSS$ Se S e S/S /SS SAlS AlSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1o} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1o} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1o k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ae /p:^(  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10rr  ]])G*G44>-?-IIST^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1rr^^ J J  66""  l l""$V%V'')@*@,,.*/*11346688:s;s==?Y@YBBD?E?GGI%J%LLN O Q~Q~SSUdVdXXZJ[J]]_0`0bbdeggiikolonnpUqUssu;v;xxz!{!}}zz^^ЉЋBB&& b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃScScSS$Sq:Sq:S~S~S[S$[S驄S驄SMSMSS$SQSQS1[S41[SST Se S e S/S /SS SAlS AlS S  S߉S߉SeS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1.bW b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1.bW b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1.bW b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ae=a9]5Y  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10xx  dd)L*L44>0?0IIST^^hhrhsh} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1xxhh X X  HH88$$   ""$h%h'')P*P,,.8/8113 4 6689;|;|==?`@`BBDDEDGGI(J(LLN O Q~Q~SSUbVbXXZF[F]]_*`*bbdeggiikdldnnpHqHssu,v,xxz{}}ff؄؆HH**  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃScS cSSSw:Sw:S~S~SOSOS梅Sp梅SSSSSSSGQS`GQSZSZSS$ Se S e SS SS S/lS /lS S  Su߉Su߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1o} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1o} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Au b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1o k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ae /p:^(  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  ]])C*C44>%?%IIST^x^xhhrXsX} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1tt`` L L  88$$  f f""$L%L'')2*2,,./11335q6q88:W;W==?9@9BBDEGGIIKnLnNNPPQPSSU2V2XXZ[]]__agbgddfIgIiik+l+nnp q s~s~uuw`x`zz|B}B$$ttTTĐĒ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃS3S 3SSSG:SG:S~S~SgSgSSpS#S#SSSQS`QSZSZSS$ S_ S _ SSS SS+S +SelS elSS S}S}S%S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ae<&wa  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10vv  hh)Z*Z44>J?JIIS6T6^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvhh Z Z  LL>>00""  "#%%'(**,,.q/q113c4c668U9U;;=G>G@@B9C9EEG+H+JJLMOOQRT~T~VVXlYl[[]Z^Z``bFcFeeg0h0jjlmooqrtstsvvxYyY{{}?~?%% {{__ёѓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS˃S˃SiS iSSS}:S}:S~S~S7S7SSpSYSYSSSAQS`AQS}ZS}ZSS$ S_ S _ S;S ;SS SMlS MlSAS ASSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1^} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1^} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1^ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ae /p:^(  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  ^^)H*H44>1?1IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qqaa Q Q  AA11 } }""$g%g'')Q*Q,,.;/;113%4%6689;;==?n@nBBDXEXGGIAJALLN)O)QQSTVVXXZm[m]]_U`Ubbd=e=ggi%j%lln o qqssuivixxzQ{Q}}99!! ii b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"b  reasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements6407LqSʃSʃS-S -SSSA:SA:S~S~SaSaSSpSSSSSQS`QS7[S7[SS$ S_ S _ S;S ;SS SMlS MlS;S ;S S SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 AeAeAeAe  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  bb)P*P44>>,,  ""$m%m'')[*[,,.I/I113747668%9%;;=>@@BCExExGGIdJdLLNNONQQS8T8VVX"Y"[[] ^ ``bbdgegggiMjMlln1o1qqstvvxxzf{f}}HH** ||^^ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS˃S˃SiS iSSS}:S}:S~S~S7S7SSpSYSYSSSQS`QSZSZSS$ S_ S _ S/S /SS SAlS AlS S  S߉S߉SwS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1_} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1_} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1_ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ae#G)v X|  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  \\)F*F44>.?.IIST^^hhrisi} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qq]] I I  55!!  k k""$U%U'')?*?,,.)/)11346688:r;r==?\@\BBDFEFGGI.J.LLNOQQSSUmVmXXZS[S]]_9`9bbdeggijlslsnnpWqWssu;v;xxz{}}uuYỶ̋??%% b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS-S -SSSA:SA:S~S~SaSaSSpSSSSSQS`QSZSZSS$ SY S Y SMS MS%S %S_lS _lSS SSS+S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ae2mW{  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10rr  bb)R*R44>@?@IIS,T,^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1rrbb R R  BB22""  "#%z%z'')j*j,,.Z/Z113J4J668:9:;;=*>*@@BCEEG H JJLLNlOlQQSXTXVVXDYD[[]0^0``bceeghj~j~llnjojqqsVtVvvxByB{{}.~.||hhޑޓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS ˃S ˃SoS oSSS}:S}:S~S~SOSOSSpSYSYSSS;QS`;QSwZSwZSS$ SY S Y S;S ;SS SMlS MlSeS eS5S5SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S! b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1w} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S! b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1w} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S! b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1w k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ae /p:^(  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  [[)I*I44>7?7IIS%T%^^h i s{s{} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mm[[ I I  77%% x x""$f%f'')T*T,,.B/B1130406689;;= > @@BBDqEqGGI_J_LLNMOMQQS;T;VVX)Y)[[]^``bce|e|ggijjjllnSoSqqs7t7vvxy{{}}qqUULJlj99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS'S 'SSS;:S;:S~S~SaSaSSpSSSSSQS`QS%[S%[SS$ SY S Y S5S 5S S  SGlS GlS;S ;S S SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ae=a9]5Y  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  \\)H*H44>2?2IIST^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp\\ H H  44    n n""$Z%Z'')F*F,,.2/21134668 9 ;;==?l@lBBDXEXGGIBJBLLN*O*QQSTVVXXZn[n]]_V`Vbbd>e>ggi&j&llnoqqssujvjxxzR{R}}::"" ~~ff b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS ˃S ˃SoS oSSS:S:S~S~SISISSpS_S_SSSQS`QSZSZSS$ SY S Y S)S )SS S;lS ;lS S  S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S# b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1v} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S# b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1v} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S# b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1v k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ae'K 1f  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  WW)A*A44>+?+IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mmWW A A  ++tt^ ^""$H%H'')2*2,,./11346{6{88:e;e==?O@OBBD9E9GGI#J#LLN O QQSSUlVlXXZV[V]]_@`@bbd*e*ggijllnnpmqmssuQvQxxz5{5}}ooSSŎŐ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS'S 'SSS5:S5:S~S~SsSsSSpSSSSSQS`QSZSZSS$ SY S Y S;S ;SS SMlS MlSqS qSASASS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1"} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1"} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1" k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Jn"FoY}  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  \\)J*J44>6?6IIS T ^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nn\\ J J  88&& y y""$g%g'')U*U,,.C/C1131416689;;= > @@BBDrErGGI^J^LLNHOHQQS2T2VVXY[[]^`{`{bbdeeeggiOjOlln9o9qqs#t#vvx y {{}}llVVˇˉ@@** b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS˃S˃SuS uSSS:S:S~S~SSSSpS_S_SSSGQS`GQSZSZSS$ SY S Y S5S 5S S  SGlS GlSkS kS;S;SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\* b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1O} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\* b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1O} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\* b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1O k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Cg5xFj8  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  aa)M*M44>7?7IIST^^his{s{} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mmaa U U  II==11%%  "#%%'(*{*{,,.g/g113S4S668?9?;;=+>+@@BCEEGHJwJwLLN_O_QQSGTGVVX/Y/[[]^``bbdsesggi[j[llnCoCqqs+t+vvxy{{}}ooWWˇˉ??'' b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS!S !SSS5:S5:S~S~SsSsSSpSSSSSQS`QS1[S1[SS$ SY S Y S#S #SS S5lS 5lS S  S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1#} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1#} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1# k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 AeAeAeAe  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  YY)C*C44>+?+IIST^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mmYY E E  11  k k""$V%V'')@*@,,.*/*11346688:s;s==?]@]BBDGEGGGI/J/LLNOQQSSUnVnXXZT[T]]_:`:bbd e ggijlylynnp_q_ssuEvExxz+{+}}jj݉݋PPÎÐ66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS˃S˃SuS uSSS:S:S~S~SSSSpSeSeSSSQS`QSZSZSS$ SY S Y S#S #SS S5lS 5lS S  S߉S߉SqS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\, b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1O} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\, b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1O} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\, b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1O k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ae+O9#t  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  ]])E*E44>+?+IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mm]] M M  ==--   ""$q%q'')Y*Y,,.A/A113)4)6689;;==?m@mBBDUEUGGI;J;LLNOQQSTVuVuXXZY[Y]]_=`=bbd!e!ggijlwlwnnp[q[ssu?v?xxz#{#}}yy]]ωϋAA%% b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS!S !SSS/:S/:S~S~SaSaSSpS S SSSQS`QS[S[SS$ SS S S S)S )SS S;lS ;lSGS GSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 u [Ae'K  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  ^^)J*J44>2?2IIST^^histst} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nnbb V V  JJ>>..  "#%|%|'')h*h,,.T/T113@4@668,9,;;=>@@BBDrErGGIZJZLLNBOBQQS*T*VVXY[[]]_n`nbbdVeVggi>j>lln&o&qqstvvxxzj{j}}RRƂƄ::xx b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS˃S˃SuS uSSS:S:S~S~SCSCSSpSeSeSSSCRS`CRS[S[SS$ SS S S SS SS S)lS )lS S  S߉S߉S_S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e3 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1t} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e3 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1t} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e3 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1t k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 * .48  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  QQ);*;44>!?!IIST^z^zhhr^s^} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **vv` `""$J%J'')4*4,,./11346}6}88:g;g==?M@MBBD3E3GGIJLLNNPrQrSSUXVXXXZ>[>]]_$`$bbd e g}g}iikclcnnpIqIssu/v/xxz{}}nnRRĉċ66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃSS SSS/:S/:S~S~S[S[SSpS S SSSYQS`YQSZSZSS$ SS S S SS SS S#lS #lS S  S߉S߉SGS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 e/S^(L  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  XX)@*@44>$?$IIST^z^zhhr\s\} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ll\\ L L  <<,, t t""$\%\'')D*D,,.,/,11346688:p;p==?T@TBBD8E8GGIJLLNOQrQrSSUVVVXXZ:[:]]_`bbdegtgtiikXlXnnp?IISS]l^lhhrLsL} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,ttZ Z""$@%@'')&*&,,. / 11335e6e88:K;K==?-@-BBDEGGIIKbLbNNPDQDSSU&V&XXZ[]y]y__a[b[ddf=g=iiklnnpqsrsruuwTxTzz|6}6hh؋؍HH b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃSS SSS):S):S~S~SgSgSSpSSSSSQS`QSZSZSS$ SM S M S5S 5S S  SGlS GlSS S_S_SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 lFjBb  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  \\)N*N44>>?>IIS*T*^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jj\\ N N  @@22$$  "#%%'')s*s,,.e/e113W4W668I9I;;=;>;@@B-C-EEGHJJL M OOQQSrTrVVX`Y`[[]N^N``b:c:eeg$h$jjlmooqqsgtgvvxMyM{{}3~3ooSSőœ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S˃S˃S{S {SSS:S:S~S~SSSSpSkSkSSSRS`RSC[SC[SS$ SM S M SS SS S/lS /lS)S )S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n< b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1O} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n< b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1O} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n< b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1O k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 {qg]  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  RR)<*<44>%?%IIS T ^^hhrisi} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%%q q""$[%['')E*E,,.///11346689;x;x==?b@bBBDLELGGI5J5LLNOQQSTVyVyXXZa[a]]_I`Ibbd1e1ggijllnoququssu]v]xxzE{E}}--ss]] b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃSS SSS#:S#:S~S~SgSgSSpSSSSS_QS`_QSZSZSS$ SM S M SS SS S/lS /lSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 q;_)j4X  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  VV)D*D44>0?0IIST^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhVV D D  22  s s""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIXJXLLNBOBQQS,T,VVXY[[]^`u`ubbd[e[ggiAjAlln%o%qqstvxvxxxzZ{Z}}<<ppRR b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S˃S˃SS SSS:S:S~S~SSSSpSkSkSSSQS`QS[S[SS$ SM S M SS SS S#lS #lS S  S߉S߉SYS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n>  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1O} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n>  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1O} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n>  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1O k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 zdNr:^  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  PP):*:44>"?"IIST^y^yhhr]s]} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))uu_ _""$I%I'')3*3,,./11346|6|88:f;f==?P@PBBD:E:GGI"J"LLNOQ{Q{SSUaVaXXZG[G]]_-`-bbdeggiikglgnnpKqKssu/v/xxz{}}iiۄۆMM33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYSʃSʃSS SSS#:S#:S~S~SUSUSSpSSSSSeQS`eQSZSZSS$ SM S M S5S 5S S  SGlS GlSS SkSkSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 i7[)l:^  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  VV)F*F44>4?4IIS T ^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&& ~ ~""$n%n'')^*^,,.N/N113>4>668.9.;;=>@@BCEEGGItJtLLN`O`QQSLTLVVX8Y8[[]$^$``bceeggirjrlln`o`qqsLtLvvx8y8{{}$~$rr^^ԑԓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYS˃S˃SS SSS:S:S~S~S7S7S Sp SqSqSSSeQS`eQSZSZSS SM S| M S)S )SQ~S Q~S/NS /NSSS SSIމSIމSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wE b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1q} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wE b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1q} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wE b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1q k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1vbNr:^  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  OO)=*=44>+?+IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaOO = =  ++~~l l""$Z%Z'')H*H,,.6/6113$4$6689;;=>@w@wBBDeEeGGISJSLLNAOAQQS/T/VVXY[[] ^ ``bbdpepggi^j^llnQoQqqs5t5vvxy{{}}ooSSŇʼn77 b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYSʃSʃSS SSS:S:S~S~SOSOSSpSSSSSRS`RS=[S=[SS$ SM S M SS SS S/lS /lS#S #S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 }si_  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP)<*<44>&?&IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((vvb b""$N%N''):*:,,.&/&11346688:t;t==?`@`BBDLELGGI6J6LLNOQQSTVzVzXXZb[b]]_J`Jbbd2e2ggijllnoqzqzssubvbxxzJ{J}}22vv^^ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYS˃S˃SS SSS:S:S~S~S7S7S Sp SqSqSSSQS`QSZSZSS SM S| M SS S?~S ?~SNS NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wG b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1p} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wG b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1p} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wG b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1p k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1'xp  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  KK)5*544>?IIS T ^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaKK 5 5    ~~hhR R""$<%<'')&*&,,./11335o6o88:Y;Y==?C@CBBD-E-GGIJLLNOQvQvSSU`V`XXZJ[J]]_4`4bbdeggijl}l}nnpiqissuMvMxxz1{1}}kk݉݋OO33 b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYSʃSʃS S  SSS:S:S~S~S[S[SSpSSSSSGQS`GQSZSZSS SG S| G S/S /SW~S W~S5NS 5NSeS eS[މS[މSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1jTxPt$H  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  PP)@*@44>,?,IIST^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGITJTLLN>O>QQS(T(VVXY[[]]_q`qbbd[e[ggiEjElln;o;qqs%t%vvxy{{}}nnXX͇͉BB,, b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYS#˃S#˃SS SSS:S:S~S~SSSSpSwSwSSSAQS`AQS}ZS}ZSS SG S| G S/S /SW~S W~S5NS 5NSeS eS[މS[މSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1O} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1O} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1O k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1v _=a?  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  UU)K*K44>5?5IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaUU I I  ==11%%  " # %%'(*y*y,,.e/e113Q4Q668=9=;;=)>)@@BCEEGHJuJuLLN]O]QQSETEVVX-Y-[[]^``bbdqeqggiYjYllnMoMqqs5t5vvxy{{}~yyaaՇՉII11 b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYSʃSʃSS SSS:S:S{~S{~SaSaSSpSSSSS RS` RSI[SI[SS SG S| G SS S?~S ?~SNS NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bv b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations15+OEi   b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  MM);*;44>#?#IIS T ^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%ss_ _""$N%N'')8*8,,."/"113 4 6688:k;k==?U@UBBD?E?GGI'J'LLN O QQSSUfVfXXZL[L]]_2`2bbdeggiikqlqnnpcqcssuIvIxxz/{/}}nnTTǎǐ:: b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYS#˃S#˃SS SSS:S:S~S~SSSSpSwSwSSSAQS`AQS}ZS}ZSS SG S| G SS S?~S ?~SNS NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1O} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1O} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1O k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1$HSwA  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  QQ)A*A44>'?'IIS T ^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaQQ A A  11!! y y""$m%m'')U*U,,.=/=113%4%668 9 ;;==?i@iBBDQEQGGI7J7LLNOQQSSUqVqXXZU[U]]_9`9bbdeggijlslsnnpcqcssuGvGxxz+{+}}ee׉׋II-- b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations1SSuʃSuʃSS|SySySSS"S|"S &SH &S9S9SK~SPK~SSXSoSoSSPSSS_S_SS$ SiS, iSS SS$ SS SEDS| EDS1RSX1RSm[S|m[SSSɅSxɅS]S\]SSS(S(SS8SSS#lS#lSySyS Sl S߉S߉S_S } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Pw b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations13 bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Pw b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations13  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Pw  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations13 bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`5eCgx6sGky9Gk|,9: b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@cq*8"F0T>bLpZ ~   h   v  !   /  ='K5YEi<BW{$Hi*N{(0T:6Z(L<`:^BfLp^ p /A/SA e  S!w!!" b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dcF) oR5{^k C J ( )    r g  W  <!kP bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1077  RR)<*<44>$?$IIST^^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations177##    !!  wwa a""$K%K'')5*5,,./113 4 6~6~88:h;h==?P@PBBD6E6GGIJLLNOQuQuSSU[V[XXZA[A]]_'`'bbdeggiikrlrnnpXqXssu>v>xxz${$}} }}ccՉՋGG++# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSS;˃S;˃SSdSSlSc!Sc!S$S$S:S@:SSSCSCSېS@ېSYS YS%S%SSSS SSS SSSd SCS CSRS RS\S \SKSKSɅSɅSSS/S/S-(S -(S#S#SSpS5lSt5lSySTyS;S;S S$ SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 C bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 C bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 C  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`X|+Ov4Lp_@dDh4X-Qx6a b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@dt  0@,P<`Lp\ l   |  (   8  $ H  4X4XDh2{ TxUyf/S -zv(Pt<*N,P@dTxh|*>.RBfVzj ~ !,!!!"@" b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1ddH,tX< hJ N  4 \  %   f  L2~d bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  UU)*44>-?-IIS!T!^^h i s}s}} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeYY M M  AA55%%  ""$u%u'')a*a,,.M/M11346688:u;u=>@@BBDmEmGGIUJULLN=O=QQS%T%VVX Y [[]]_i`ibbd]e]ggiEjElln-o-qqstvvxxzq{q}}YY͂̈́AA((  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1SSoʃSoʃSS|SySySSS"S|"S &SH &S9S9SK~SPK~SSXSoSoSSPSSS_S_SS$ SiS, iSS SS$ SS SEDS| EDSSSXSS\S|\SASASeՅSxeՅS! S\! S S S+S+SS8SSSlSlSySyS Sl SQ߉SQ߉SS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ot b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations13 bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ot b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations13  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ot  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations13 bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`BpLp(Fj),Pnu&Tx-Q *N|  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Lp(t,PTx 0|4 X    \  8   <`d@Dh l$HHl"HljNr DTxZ~Bf`f dl >  r  !!<"`""###^$$%8%%%&&6'Z''(((X)|)*2***z++0,T,, --- b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d(<Pdx  4 H Z l ~ F %"c4BFXj| bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1088  PP)6*644>?IIST^y^yhhr[s[} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations188$$    ##ssY Y""$?%?'')%*%,,. / 1~1~335d6d88:J;J==?.@.BBDEGGIIKcLcNNPEQESSU'V'XXZ [ ]z]z__ahbhddfJgJiik,l,nnpqssuuwaxazz|C}C%%wwWWǐǒ# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSS;˃S;˃SSdSSlSc!Sc!S$S$S:S@:SSSCSCSېS@ېSYS YS+S+SSSS SSS SSSd SCS CS?SS ?SS{\S {\SÛSÛS!΅S!΅SSSSS)S )S S SSpSlStlSySTyS S S߉S$߉SAS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p< b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 C bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p< b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 C bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p<  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 C  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`;Dh.R<X|.RY})M(/S{$H b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@w6Hl~=Os DVz  ' K  ]    . R  d#5Yk*<`n+(q9]Kok%I(#l6ZFjh D%3We"0Tb  !!!-"Q"""_###$$$*%N%%%\&& b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dwne\SJA8/&    n  z 9  oKdYNC8 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  OO)*44>!?!IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33##k k""$W%W'')?*?,,.'/'1t1t335\6\88:D;D==?Q@QBBD5E5GGIJLLNNPoQoSSUSVSXXZ7[7]]_`bbd e g}g}iikalannpEqEssu)v)xxz { }}ccՄՆGG))   b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSʃSʃS Sd SSlSi!Si!S$S$S:S@:S{~S{~SISISᐄS@ᐄS_S _SSSSSS SYS YSSd SCS CSQS QSZS ZS S SsDžSsDžS/S/SiSiSg'S g'SSSSpS/lSt/lSySTyS#S#S߉S$߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 (C bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 (C bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 (C  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`4DhEi{8=Mq&CyTx&Jg b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Z~VzRvNrJnFjBf>b:^6 Z  2 V  . R  * N  & J  "FB@Rv@Tx@Vz@X|AZ~C\!E^#G`%I%I%I%I%I%I%I%I b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dZ4hBvP*` V  < 3    t m P J , d@ bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  MM)7*744>>HIST^^hhrrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``PP @ @  00    l l""$V%V'')@*@,,.*/*11346688:s;s==?]@]BBDGEGGGI J LLNNPhQhSSUvVvXXZ^[^]]_F`Fbbd8e8ggi j llnoq|q|ssudvdxxzL{L}}44||ff  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSS5˃S5˃SSdSSlSo!So!S$S$S:S@:S S SOSOS琄S@琄SeS eSSSSSS S_S _SSd S%CS %CSRS RS3\S 3\S{S{SDžSDžSeSeSSS'S 'S)S)SSpS;lSt;lSySTySSSSS$SSS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a- b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 FU  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a- b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 FU  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a-  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 FU   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Cg0T|5@xp4Fj\8 Pt<` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@[Y}W{UySwQuOsMqKoI m  G k  E i  C g  A e  &?cq=aOs=a-Q?c /z AeX|Cg6ZEi:GkImKoMqOsQuSwUyW{Y}[  b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d[6lG"}X3k ]  H (  % s >   _ < vS0 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  WW)I*I44>;?;IIS'T'^^hhrs} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeWW I I  ;;--  "#%|%|'')n*n,,.`/`113R4R668D9D;;=6>6@@B(C(EEGHJJL M OOQQSoToVVX]Y][[]K^K``b7c7eeg!h!jjl m ooqqsftfvvxLyL{{} ~ eeppTTƑƓ  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSʃSʃSSdSSlSi!Si!S$S$S:S@:S{~S{~SISISᐄS@ᐄS_S _SSSSSS SYS YSSd SCS CSRS RS\S \SKSKSDžSDžSqSqScScSa'S a'S S SSpSlStlSySTyS S S߉S$߉SYS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 'C bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 'C bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 'C  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`!Eqv,4Uy3WDh]#Gx6W{ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@[Y}W{UySwQuOsMqKoI m  G k  E i  C g  A e  ?c=a9]3W-Q'K!E?93 -b'd! zf th njhl bn\p  b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d[6lG"}X3g  @   r K $ '  }Va/> bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  KK)5*544>>HIS T ^~^~hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``LL 8 8  $$ppZ Z""$D%D'').*.,,./11346w6w88:a;a==?K@KBBD5E5GGIIKkLkNNPQQQSSU^V^XXZD[D]]_*`*bbdeggiiklllnnpPqPssu4v4xxz{}}nnTTljNj::   b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSS5˃S5˃SSdSSlSo!So!S$S$S:S@:SSSOSOS琄S@琄SeS eS%S%SSSS S_S _SSd S%CS %CS1RS 1RSm[S m[SSS;ÅS;ÅSSSSS%S %SSSSpS)lSt)lSySTySSS߉S$߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e1 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 GT  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e1 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 GT  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e1  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 GT   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`d<Dhp:Sw %I|Qu&#G2 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Cg5xFj8{Im;~ Lp> Os A    R v   D    U y  # G   X|&J[)M^,Pa/Sc!Aed2VC$!gl5YJn(L*u b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dCI OU[a$ g *  m  0 s T 6  yj<5 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  QQ)?*?44>-?-IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccQQ ? ?  --  n n""$\%\'')J*J,,.8/8113&4&6689;;=>@y@yBBDgEgGGIUJULLN?O?QQS)T)VVXY[[]]_r`rbbdXeXggi>j>lln&o&qqstvyvyxxz4{4}}ssUU  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLSSTS{!S{!S$S$S:S(:S~S~S[S[SS(SqSqSSSSSS SkS kSSL S1CS 1CS[RS [RS[S [SߚSߚS?S?SSSSDSSS/lS /lSGSGSS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 C bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 C bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 C  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`3CgGkz AdPt(L`8"6Z4X b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@)'%#~!|zxv bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations13  i D   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  NN)<*<44>,?,IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``NN < <  **}}k k""$Y%Y'')G*G,,.5/5113#4#6689;;==?v@vBBDdEdGGITJTLLNBOBQQS0T0VVXY[[] ^ ``bbdqeqggi_j_llnHoHqqs,t,vvxy{{}}ff؂؄JJ.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS)˃S)˃SSLSSTS!S!S$S$S:S(:SSSaSaSS(SwSwSSS}S}SS SqS qSSL S7CS 7CS-SS -SSi\S i\SSSSSSS/SD/SSSAlS AlSSSeS0eS S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V" b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Qq bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V" b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Qq bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V"  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Qq  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Cg5Y9](L/S'Kt%I>s=f 0 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@9]?!nPt2V8g bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1h 5  O  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  UU)E*E44>5?5IIS!T!^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%% } }""$m%m'')]*],,.M/M113=4=668-9-;;=>@@B C EEGGIuJuLLNaOaQQSMTMVVX9Y9[[]%^%``bceeggisjslln_o_qqsKtKvvx7y7{{}#~#qq]]ӑӓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃS SL SSTS{!S{!S$S$S:S(:S~S~S[S[SS(SqSqSSSSSS SkS kSSL S1CS 1CSRS RS9\S 9\SSS1DžS1DžS3S3S SD SSSlS lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 C bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 C bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 C  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$HLp|,P@d 0(\| gDh<` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@trpnlj h f bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 z U0 f } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  JJ)4*444> ? IIS T ^|^|hhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``JJ 4 4  }}ggQ Q""$;%;'')%*%,,./11335n6n88:X;X==?B@BBBD,E,GGIJLLNOQwQwSSUaVaXXZK[K]]_5`5bbdeggi j l~l~nnpbqbssuFvFxxz*{*}}dd։֋HH,, b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS/˃S/˃SSLSSTS!S!S$S$S:S(:S S SaSaSS(SwSwSSSSSS SqS qSSL S7CS 7CSeQS eQSZS ZS际S际SÅSÅSSSSDSSS)lS )lSSS߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z& b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Rr bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z& b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Rr bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z&  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Rr  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Cg=Im5$Hxa"Fj;_k:9Ei b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Im+O 1~`Bf$H*w  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 i 6P } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  OO);*;44>'?'IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''uua a""$M%M'')9*9,,.%/%11346688:s;s==?_@_BBDKEKGGI7J7LLNOQQSTV{V{XXZc[c]]_K`Kbbd3e3ggijllnoqwqwssu_v_xxzG{G}}//ss[[ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTS!S!S %S %S#:S(#:S~S~SmSmSS(SSSSSSSS S}S }SSL SCCS CCSRS RS[S [S'S'SپSپSSS#SD#SSS5lS 5lSYSYS)S0)SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 E  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 E  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 E   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Cg2Im{ #GwUy5!/Sj -EiDhy b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@5Y/S)M#GA;5 / bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 k D  v  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  XX)D*D44>0?0IIST^^histst} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddXX L L  @@44((  "#%%'')r*r,,.^/^113J4J668696;;=">"@@BCEEGGIpJpLLNXOXQQS@T@VVX(Y([[]^``bbdlelggiTjTllnbx 8\Z~Bf<` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@pVz<`"F,{aGk-Q bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1&u D  b  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>/?/IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIWJWLLNAOAQQS+T+VVXY[[]]_t`tbbd^e^ggiHjHlln2o2qqstvvxy{{{{}}eeڂڄOOćĉ99## b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTS!S!S%S%S#:S(#:S~S~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CS3SS 3SSo\S o\SSSƅSƅSSSSDSSS#lS #lS S S߉S0߉S_S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 D bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 D bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 D  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Cg'K&J(Ls8=t'>ye /f 0 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@)#|vpj d^ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 O (Z3 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  TT)<*<44>$?$IIST^z^zhhr^s^} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$ | |""$h%h'')P*P,,.8/8113 4 6689;|;|==?d@dBBDLELGGI4J4LLNOQQSSUnVnXXZR[R]]_6`6bbdeggiikplpnnpTqTssu8v8xxz{}}rrVVȉȋ:: b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S)˃S)˃SSLSSTS!S!S%S%S:S(:SSSySySS(SSSSS}S}SS SS SSL SOCS OCSQS QSZS ZSSS+ąS+ąS=S=SSDSSS#lS #lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ]M0 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ]M0 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}O  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ]M0  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Cg>Os5$5Yys Gk?Y}9%?ct b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@9nTx:^ D*y_ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations11 Om } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP):*:44>$?$IIS T ^}^}hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((vvb b""$M%M'')7*7,,.!/!113 4 6688:j;j==?T@TBBD>E>GGI(J(LLNOQQSSUgVgXXZM[M]]_3`3bbdeggiikrlrnnpXqXssu>v>xxz${$}} }}cc։֋II// b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTS!S!S/%S/%S):S():S~S~SSS)S()SSSSSSS S  SS SSL SgCS gCSQS QSZS ZS S SsSsSSSSDSSS#lS #lS S S߉S0߉S_S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  [@ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  [@ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  [@  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` 1hKo|?c)MZ~;+8\v :[Tx b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@k aW{MqCg9]/S%I bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1qH w N  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  LL)6*644>?IIST^w^whhr]s]} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%qq[ [""$E%E'')/*/,,./11346x6x88:b;b==?J@JBBD0E0GGIJLLNNPoQoSSUUVUXXZ;[;]]_!`!bbdegzgziik`l`nnpFqFssu,v,xxz{}}kkބކQQÉË55 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S˃S˃SSLSSTS!S!S%S%S:S(:S~S~SsSsS S( SSS S SqSqSS SS SSL SICS ICSuSS uSS\S \SSSSSSSMSMS)SD)SSS;lS ;lSASASS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p@  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 \f" bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p@  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 \f" bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p@   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 \f"  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`;_ "8\u EieRvBfOs!Lp b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@*{eOs9]#G 1 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations11 S $ u  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  YY)E*E44>/?/IIST^^hhrsss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ii]] Q Q  EE99))  "#%w%w'')c*c,,.O/O113;4;668'9';;=>@@BBDoEoGGIWJWLLN?O?QQS'T'VVXY[[]]_k`kbbdSeSggi;j;lln#o#qqs t vvxxzg{g}}OOÂÄ77ww b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTS!S!S)%S)%S):S():S~S~SSS#S(#SSSSSSSS SS SSL SaCS aCScSS cSS\S \S盅S盅S#ƅS#ƅSSSSDSSSlS lS S SQ߉S0Q߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  Z> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  Z> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  Z>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`*N^4Xb.Rf 8\j2V<` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@?5+!xn dZ~Pt bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 % | S* } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  LL)2*244>?IISS]i^ihhrKsK} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''ooU U""$;%;'')!*!,,./1z1z335`6`88:F;F==?*@*BBD E G}G}IIK_L_NNPAQASSU#V#XXZ[]v]v__aXbXddf:g:iiklnnpprosouuwQxQzz|3}3gg׋׍GG b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S#˃S#˃SSLSSTS!S!S%S%S:S(:S~S~SsSsS S( SSS S SqSqSS SS SSL SICS ICSQS QS[S [SgSgSąSąSaSaSSDSSS#lS #lS S S߉S0߉SAS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1rD b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ]g$ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1rD b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ]g$ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1rD  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ]g$  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`?wQu)e;_zSw%IdAe3Nr b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@lVz@d*N8"s ] bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1F  h9  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  SS);*;44>!?!IIST^w^whhr[s[} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggWW G G  77''o o""$W%W'')?*?,,.'/'11346688:k;k==?Q@QBBD5E5GGIJLLNNPoQoSSUSVSXXZ7[7]]_`bbddfqgqiikUlUnnp9q9ssuvxxz{}s}sWWɄɆ;; b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1peSʃSʃSSTSS\S{!S{!S$S$S/:S0/:S~S~S[S[SS0SqSqSSS S SS SkS kSST S1CS 1CSIS ISPSH PSYS YS)S )S+S+SSS#SX#SSS5lS 5lS/S/S߉SD߉SSH} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 C  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 C  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 C   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)")B)Z-b1reasonLocalCopyvalue7bytes144patternContiguousinvocations8k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  .c;p"F{ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@"F<2(uk aW{ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1R) W .  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  QQ);*;44>&?&IIST^^hhrlsl} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$p p""$Z%Z'')D*D,,../.11346689;w;w==?a@aBBDKEKGGI6J6LLNOQQSTVzVzXXZb[b]]_J`Jbbd2e2ggijllnoqvqvssu^v^xxzF{F}}..vv``! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1" bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS˃S˃S{SL{SSTS{!S{!S$S$S:S(:S~S~S[S[SS(SqSqSSSkSkSS SkS kSSL S1CS 1CS SS  SSE\S E\SSSOSOSSS;SD;SSSMlS MlSSSeS0eS S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e5 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 EQ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e5 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 EQ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e5  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 EQ  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`4X|5@t=8\ !ETx)M<` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@`RvDh6Z(L> 0"w bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1f;  e :  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  [[)M*M44>???IIS+T+^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ii[[ M M  ??11##  "#%%'')r*r,,.d/d113V4V668H9H;;=:>:@@B,C,EEGHJJLMOOQQSsTsVVXaYa[[]O^O``b;c;eeg%h%jjlmooqqshthvvxNyN{{}4~4rrVVȑȓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSSLSSTS{!S{!S$S$S):S():S~S~S[S[SS(SqSqSSSSSS SkS kSSL S1CS 1CSiSS iSS\S \S훅S훅SŅSŅSSSSDSSS)lS )lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 C  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 C  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 C   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`-QCg56ZzHl3W::^ 7[>b b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@MqCg9]/S%I?5+ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1  \ 3 a } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  OO)9*944>#?#IIST^z^zhhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((tt^ ^""$H%H'')2*2,,./11346{6{88:e;e==?O@OBBD9E9GGI#J#LLN O Q|Q|SSUbVbXXZH[H]]_.`.bbdeggiikhlhnnpLqLssu0v0xxz{}}jj܄܆PPÉË66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS˃S˃SSLSSTS{!S{!S$S$S:S(:S~S~S[S[SS(SqSqSSSkSkSS SkS kSSL S1CS 1CSRS RSO[S O[SSSŅSŅSSS#SD#SSS5lS 5lS#S#S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g9 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 DR bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g9 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 DR bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g9  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 DR  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`8Cg{&5Im}x;qFj~ 8e b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@i[Mq?c1U#G9+ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1  d 9c } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  UU)C*C44>1?1IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggUU C C  11  r r""$`%`'')N*N,,.@}@}BBDkEkGGIYJYLLNCOCQQS-T-VVXY[[]^`v`vbbd\e\ggiBjBlln&o&qqstvyvyxxz[{[}}==ssUU b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS!SL!SSTS!S!S%S%S5:S(5:S~S~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CSwQS wQSZS ZSSS5S5S%S%S/SD/SSSAlS AlSYSYS)S0)SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 D  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 D  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H|  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 D   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`+nAek<`7=a| . 3qNrx?c"F b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@*6B*N6ZBfNrZ~ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1@ "  f H  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  RR)@*@44>0?0IIST^^histst} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddRR @ @  ..  o o""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGIXJXLLNFOFQQS4T4VVX"Y"[[]^``bbdueuggicjcllnLoLqqs0t0vvxy{{}}jj܂܄NN22 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS˃S˃SuSLuSSTS!S!S$S$S:S(:S~S~SaSaSS(SwSwSSS_S_SS SqS qSSL S7CS 7CS[RS [RS[S [SߚSߚSYSYS1S1SASDASSSSlS SlSSS}S0}S%S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z* b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 H ] bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z* b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 H ] bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z*  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 H ]  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`)Mf&=a|z(Quq<ee,P b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@DhLpTx\d lt| bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1T 4  t T  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  YY)I*I44>9?9IIS%T%^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiYY I I  99))   ""$q%q'')a*a,,.Q/Q113A4A668191;;=!>!@@BCEEGHJyJyLLNeOeQQSQTQVVX=Y=[[])^)``bceeghjwjwllncocqqsOtOvvx;y;{{}'~'uuaaבד b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS!SL!SSTS!S!S%S%S/:S(/:S~S~SgSgSS(S}S}SSS S SS SwS wSSL S=CS =CSSS SSW\S W\SSSȅSȅSSS#SD#SSS5lS 5lSSS߉S0߉S}S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 D bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 D bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H|  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 D  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Cg&J~5 Dxj 0Fjd8*PtJn b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@fr~(4@(L4 X  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1* nP2 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  NN)8*844>$?$IIST^^hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddNN 8 8  ""  kkU U""$?%?''))*),,./11335r6r88:\;\==?F@FBBD0E0GGIJLLNOQ{Q{SSUeVeXXZO[O]]_9`9bbd#e#ggi j llnnpfqfssuJvJxxz.{.}}hhډڋLL00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS˃S˃SuSLuSSTS!S!S$S$S:S(:S~S~SaSaSS(SwSwSSSeSeSS SqS qSSL S7CS 7CSRS RSC[S C[SSSȅSȅSSS)SD)SSS;lS ;lS5S5SS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\. b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 G ] bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\. b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 G ] bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\.  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 G ]  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` -1Uj#(LwTx :gEiwW{ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$,4< D(L0T8 \  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations14 tT4 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  SS)?*?44>+?+IIST^^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++yye e""$Q%Q'')=*=,,.)/)11346689;w;w==?c@cBBDOEOGGI;J;LLN#O#QQS T VVXXZg[g]]_O`Obbd7e7ggijllnoq{q{ssucvcxxzK{K}}33ww__ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS'SL'SSTS!S!S %S %S5:S(5:S~S~SmSmSS(SSSSSSSS S}S }SSL SCCS CCSqQS qQSZS ZSSS?…S?…SSS;SD;SSSMlS MlSqSqSAS0ASS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 C bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 C bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I}  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 C  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$g1U^5Y++Or'%jHll8\C b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@r'%IGki@>b` bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1/  v c  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  \\)H*H44>4?4IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hh\\ P P  DD88,,   "#%%'(*v*v,,.b/b113N4N668:9:;;=&>&@@BCEEGGItJtLLN\O\QQSDTDVVX,Y,[[]^``bbdpepggiXjXlln@o@qqs(t(vvxy{{}}llTTȇȉ<<$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS ˃S ˃SoSLoSSTS!S!S%S%S}:S(}:S~S~SgSgSS(S}S}SSSYSYSS SwS wSSL S=CS =CSQS QS7[S 7[SSS]…S]…SSS;SD;SSSMlS MlSqSqSAS0ASS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 K\ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 K\ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 K\  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`BPt<oJn 6]Dh 0Ko>b b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ D>b\z -'KEic bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1B -   n  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  WW)E*E44>3?3IIST^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiWW E E  33!!t t""$b%b'')P*P,,.>/>113,4,6689;;=>@@BBDmEmGGI[J[LLNEOEQQS/T/VVXY[[]^`x`xbbdbebggiLjLlln6o6qqs t vvx y {{}}iiނބSSȇȉ=='' b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS!SL!SSTS!S!S %S %S5:S(5:S~S~SmSmSS(SSSSSSSS S}S }SSL SCCS CCSsRS sRS[S [SSSQ˅SQ˅SeSeS#SD#SSS5lS 5lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 C bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 C bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I}  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 C  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`CgAh85Y{ 0ImMq;$He<` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@75YW{ y . ,!P!!!N"r""#p## bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 P=* } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  XX)@*@44>(?(IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhXX H H  88((  ""$l%l'')T*T,,.>"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS ˃S ˃SoSLoSSTS!S!S%S%S:S(:S~S~SgSgSS(S}S}SSS_S_SS SwS wSSL S=CS =CSRS RSU[S U[SSSQ˅SQ˅SeSeS)SD)SSS;lS ;lS S S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q# b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 J ] bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q# b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 J ] bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q#  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 J ]  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H_&A|Vz|"FpFjd b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@4.RLpj ; 5!Y!!!S"w""#q## bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 YD/ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  TT)>*>44>(?(IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,zzf f""$Q%Q'');*;,,.%/%11346688:n;n==?X@XBBDBEBGGI,J,LLNOQQSSUkVkXXZQ[Q]]_7`7bbdeggijlvlvnnp\q\ssuBvBxxz({(}}ggډڋMM33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃSWSWSS$Sk:Sk:S~S~SaS$aS㩄S㩄SGSGSS$SQSQS[S4[SST Se S e S5S 5S S  SGlS GlS S  S߉S߉SS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1)][  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1)][  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1)][  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Cg9]-Q!E  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10vv  aa)K*K44>1?1IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvbb N N  ::&&p p""$Z%Z'')D*D,,../.11346689;w;w==?]@]BBDCECGGI)J)LLNOQQSSUhVhXXZN[N]]_4`4bbdeggijlslsnnpYqYssu?v?xxz%{%}} ~~bbԉԋFF** b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS˃S˃SiS iSSS}:S}:S~S~SISISSpSYSYSSSYQS`YQSZSZSS$ Se S e SGS GSS SYlS YlS_S _S/S/SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1k} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1k} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1k k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Cg5xFj9  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10vv  ff)R*R44>:?:IIS"T"^^h i s|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvjj ^ ^  RRFF66""  "#%%'')p*p,,.\/\113H4H668494;;=>@@BCEzEzGGIbJbLLNJOJQQS2T2VVXY[[]^`v`vbbd^e^ggiFjFlln.o.qqstvvxxzr{r}}ZZ΂΄BB''  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃS]S]SS$Sq:Sq:S~S~SaS$aS驄S驄SMSMSS$SRSRSC[S4C[SST Se S e S#S #SS S5lS 5lS S  Su߉Su߉SS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1*^Y  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1*^Y  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1*^Y  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 CgCgCgCg  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10vv  __)E*E44>'?'IIS T ^z^zhhrZsZ} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvbb N N  ::&&h h""$N%N'')4*4,,./11346s6s88:Y;Y==?;@;BBDEGGIIKpLpNNPRQRSSU4V4XXZ[]]__aibiddfKgKiik-l-nnpqssuuwbxbzz|D}D&&vvVVƐƒ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS˃S˃SiS iSSS}:S}:S~S~SISISSpSYSYSSSeQS`eQSZSZSS$ Se S e S/S /SS SAlS AlS S  S߉S߉S_S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1j} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1j} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Cw b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1j k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Cg9~ Pt"F  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10vv  bb)J*J44>.?.IIST^^hhrfsf} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvff V V  FF66""  ~ ~""$f%f'')N*N,,.6/611346689;z;z==?^@^BBDBEBGGI&J&LLN O Q|Q|SSU`V`XXZD[D]]_(`(bbd e g~g~iikblbnnpFqFssu*v*xxz{}}ddքֆFF((  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃS9S 9SSSM:SM:S~S~SaSaSũSpũS)S)SSSQS`QSZSZSS$ S_ S _ S;S ;SS SMlS MlSGS GSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Cg&J8&y  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10uu  bb)L*L44>5?5IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1uuee U U  EE55""   ""$k%k'')U*U,,.?/?113)4)6689;;==?r@rBBD\E\GGIEJELLN-O-QQSTVVXXZq[q]]_Y`YbbdAeAggi)j)llnoqqssumvmxxzU{U}}==%%mm b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS ˃S ˃SoS oSSS}:S}:S~S~S7S7SSpSYSYSSSSQS`SQSZSZSS$ S_ S _ SMS MS%S %S_lS _lSS SwSwSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1\} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1\} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1\ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Cg5xFj8  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10rr  dd)V*V44>F?FIIS2T2^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1rrdd V V  HH::,,  "#%%'(*{*{,,.m/m113_4_668Q9Q;;=C>C@@B5C5EEG'H'JJLMOOQRTzTzVVXhYh[[]V^V``bBcBeeg,h,jjlmooqqsotovvxUyU{{};~;!!ww[[͓͑ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃS3S 3SSSG:SG:S~S~SaSaSSpS#S#SSSIRS`IRS[S[SS$ S_ S _ S/S /SS SAlS AlSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 CgW{k -  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  ^^)H*H44>0?0IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ss__ K K  77##m m""$W%W'')A*A,,.+/+11346688:t;t==?^@^BBDHEHGGI0J0LLNOQQSSUoVoXXZU[U]]_;`;bbdeggijlulunnpYqYssu=v=xxz!{!}}ww[[Ή΋AA'' b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS ˃S ˃SoS oSSS:S:S~S~S7S7SSpS_S_SSSQS`QSZSZSS$ S_ S _ S;S ;SS SMlS MlS5S 5SSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1]} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1]} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1] k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Cg!E#nNr  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10rr  ``)N*N44>:?:IIS"T"^^histst} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1rr`` N N  <<** } }""$k%k'')Y*Y,,.G/G113545668#9#;;=>@@BBDvEvGGIbJbLLNLOLQQS6T6VVX Y [[] ^ ``bbdeeeggiKjKlln/o/qqstvvxxzd{d}}FF(( zz\\ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS3S 3SSSA:SA:S~S~SaSaSSpSSSSSQS`QSZSZSS$ S_ S _ SAS ASS SSlS SlSkS kS;S;SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Cg@ .o  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  __)M*M44>;?;IIS)T)^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qq__ M M  ;;)) | |""$j%j'')X*X,,.F/F113444668"9";;=>@@BBDuEuGGIcJcLLNQOQQQS?T?VVX-Y-[[]^``b c eegginjnllnWoWqqs;t;vvxy{{}~uuYYˇˉ==!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS˃S˃SuS uSSS:S:S~S~SISISSpS_S_SSSMQS`MQSZSZSS$ S_ S _ SMS MS%S %S_lS _lSS SSS+S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U% b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1q} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U% b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1q} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U% b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1q k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Cg5xFj8  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  ^^)N*N44>>..  ""$v%v'')f*f,,.V/V113F4F668696;;=&>&@@BCEEGHJ|J|LLNhOhQQSTTTVVX@Y@[[],^,``bceeghjzjzllnfofqqsRtRvvx>y>{{}*~*xxddڑړ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS-S -SSSA:SA:S~S~SaSaSSpSSSSS7RS`7RSs[Ss[SS$ S_ S _ S/S /SS SAlS AlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 CgSwcs  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10oo  YY)C*C44>-?-IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ooYY C C  --vv` `""$J%J'')4*4,,./11346}6}88:g;g==?Q@QBBD;E;GGI%J%LLNOQQSSUnVnXXZX[X]]_B`Bbbd,e,ggijllnnpoqossuSvSxxz7{7}}qqUUǎǐ99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS˃S˃SuS uSSS:S:S~S~SISISSpSeSeSSSwQS`wQSZSZSS$ S_ S _ S5S 5S S  SGlS GlS;S ;S S SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U' b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1p} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U' b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1p} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U' b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1p k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 CgBe?c  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  ZZ)F*F44>0?0IIST^^histst} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nnZZ F F  22  l l""$X%X'')D*D,,.0/011346689;~;~==?j@jBBDVEVGGI@J@LLN(O(QQSTVVXXZl[l]]_T`Tbbd;?;IIS#T#^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qqee Y Y  MMAA55))  "#%%' ( **,,.k/k113W4W668C9C;;=/>/@@BCEEGHJ{J{LLNcOcQQSKTKVVX3Y3[[]^``bcewewggi_j_llnGoGqqs/t/vvxy{{}}ss[[χωCC++ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS˃S˃SuS uSSS:S:S~S~SSSSpSeSeSSSeQS`eQSZSZSS$ SY S Y S5S 5S S  SGlS GlSqS qSASASS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^. b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1M} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^. b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1M} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^. b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1M k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Hl@Z~.R  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  XX)F*F44>2?2IIST^^his{s{} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjXX F F  44""u u""$c%c'')Q*Q,,.?/?113-4-6689;;= > @@BBDnEnGGIZJZLLNDODQQS.T.VVXY[[]^`w`wbbdaeaggiKjKlln5o5qqstvvx y {~{~}}hh݂݄RRLJlj<<&& b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS'S 'SSS;:S;:S~S~SsSsSSpSSSSS=RS`=RSy[Sy[SS$ SY S Y S#S #SS S5lS 5lS S  S߉S߉SqS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 CgW{k -  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10oo  __)G*G44>-?-IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1oo__ O O  ??//  ""$s%s'')[*[,,.C/C113+4+6689;;==?o@oBBDWEWGGI=J=LLN!O!QQSTVwVwXXZ[[[]]_?`?bbd#e#ggijlylynnp]q]ssuAvAxxz%{%}} {{__щыCC'' b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS˃S˃S{S {SSS:S:S~S~SSSSpSeSeSSSQS`QSZSZSS$ SY S Y S#S #SS S5lS 5lSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^0 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1M} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^0 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1M} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^0 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1M k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Cg!E#nLp  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  WW)A*A44>)?)IIST^^hhrhsh} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kkWW C C  //}}i i""$T%T'')>*>,,.(/(11346688:q;q==?[@[BBDEEEGGI-J-LLNOQQSSUlVlXXZR[R]]_8`8bbdeggijlwlwnnp]q]ssuCvCxxz){)}}hhۉۋNN44 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS!S !SSS5:S5:S~S~SaSaSSpSSSSSRS`RSO[SO[SS$ SS S S SS SS S/lS /lS S  S߉S߉SeS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ~vnf  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  UU)?*?44>%?%IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjVV B B  ..zzd d""$N%N'')8*8,,."/"113 4 6688:k;k==?Q@QBBD7E7GGIJLLNOQvQvSSU\V\XXZB[B]]_(`(bbdeggiikglgnnpMqMssu3v3xxz{}}rrVVȉȋ:: b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS˃S˃S{S {SSS:S:S~S~S=S=SSpSkSkSSSQS`QS7[S7[SS$ SS S S S)S )SS S;lS ;lSAS ASSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g7 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1n} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g7 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1n} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g7 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1n k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 {m `Rv  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  ZZ)F*F44>.?.IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jj^^ R R  FF::**  "#%x%x'')d*d,,.P/P113<4<668(9(;;=>@@BBDnEnGGIVJVLLN>O>QQS&T&VVXY[[]]_j`jbbdReRggi:j:lln"o"qqs t v~v~xxzf{f}}NN‚„66tt b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS!S !SSS/:S/:S~S~SaSaSSpS S SSSQS`QSZSZSS$ SS S S SS SS SlS lS S  SW߉SW߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 kAe;X|  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  SS)9*944>?IISS]n^nhhrNsN} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjVV B B  ..vv\ \""$B%B'')(*(,,./11335g6g88:M;M==?/@/BBDEGGIIKdLdNNPFQFSSU(V(XXZ [ ]{]{__a]b]ddf?g?iik!l!nnpqststuuwVxVzz|8}8jjڋڍJJ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS˃S˃SS SSS:S:S~S~S=S=SSpSkSkSSSQS`QS+[S+[SS$ SS S S SS SS S#lS #lS S  S߉S߉SGS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g9 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1m} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g9 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1m} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g9 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1m k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 |jX|Fj  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  VV)>*>44>"?"IIST^x^xhhrZsZ} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjZZ J J  ::**r r""$Z%Z'')B*B,,.*/*11346688:n;n==?R@RBBD6E6GGIJLLNNPpQpSSUTVTXXZ8[8]]_`bbdegrgriikVlVnnp:q:ssuvxxz{}t}tXXʄʆ:: b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃSS SSS/:S/:S~S~SgSgSSpS S SSSQS`QS[S[SS$ SS S S S#S #SS S5lS 5lS/S /S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 u_Im3W  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  VV)@*@44>)?)IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiYY I I  99)) u u""$_%_'')I*I,,.3/311346689;|;|==?f@fBBDPEPGGI9J9LLN!O!QQS T V}V}XXZe[e]]_M`Mbbd5e5ggijllnoqyqyssuavaxxzI{I}}11wwaa b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S˃S˃SS SSS:S:S~S~SSS Sp SqSqSSSQS`QS[S[SS$ SS S S S/S /SS SAlS AlSS S_S_SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p@  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1M} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p@  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1M} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p@  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1M k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 weSwAe  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  XX)J*J44>:?:IIS&T&^^h i s}s}} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffXX J J  <<..    "#%}%}'')o*o,,.a/a113S4S668E9E;;=7>7@@B)C)EEGHJJL M OOQQSnTnVVX\Y\[[]J^J``b6c6eeg h jjl m o}o}qqsctcvvxIyI{{}/~/kk݌ݎOO b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃSS SSS):S):S~S~SgSgSSpSSSSSeQS`eQSZSZSS$ SS S S SS SS S#lS #lS S  S߉S߉S_S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 g5Y'j8\  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  RR)<*<44>$?$IIS T ^{^{hhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++wwa a""$K%K'')5*5,,./113 4 6~6~88:h;h==?R@RBBD.?.IIST^^hhrhsh} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffTT B B  00  q q""$_%_'')M*M,,.;/;113)4)6689;;=>@|@|BBDjEjGGIVJVLLN@O@QQS*T*VVXY[[]]_s`sbbdYeYggi?j?lln#o#qqstvvvvxxzX{X}}::nnߎߐPP b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃSS SSS):S):S~S~SUSUSSpSSSSSQS`QSZSZSS$ SM S M S#S #SS S5lS 5lSSS SS#S#SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 pPt0T4  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>/?/IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIWJWLLNEOEQQS3T3VVX!Y![[]^``bbdtetggibjbllnMoMqqs1t1vvxy{{}}kk݂݄OO33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S#˃S#˃SS SSS:S:S~S~S1S1SSpSwSwSSSQS`QSZSZSS SM S| M S;S ;Sc~S c~SANS ANSS SމSމS?S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yI b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1k} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yI b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1k} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yI b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1k k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1w3!tb  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  RR)B*B44>0?0IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbRR B B  22"" z z""$j%j'')Z*Z,,.J/J113:4:668*9*;;=>@@B C EEGGIpJpLLN\O\QQSHTHVVX4Y4[[] ^ ``b c eegginjnllndodqqsPtPvvx!?!IIS T ^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccMM 7 7  !!  jjT T""$>%>'')(*(,,./11335q6q88:[;[==?E@EBBD/E/GGIJLLNOQxQxSSUbVbXXZL[L]]_6`6bbd e ggi j llnnpgqgssuKvKxxz/{/}}iiۉۋMM11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S#˃S#˃SS SSS:S:S~S~S1S1SSpSwSwSSS+RS`+RSg[Sg[SS SM S| M S#S #SK~S K~S)NS )NS)S )SމSމSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yK b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1j} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yK b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1j} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yK b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1j k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1&JJnn$  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  NN):*:44>$?$IIS T ^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&tt` `""$L%L'')8*8,,.$/$11346688:r;r==?^@^BBDJEJGGI4J4LLNOQQSTVxVxXXZ`[`]]_H`Hbbd0e0ggijllnoq|q|ssudvdxxzL{L}}44xx`` b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYSʃSʃSS SSS#:S#:S~S~S[S[SSpSSSSSMQS`MQSZSZSS SG S| G S/S /SW~S W~S5NS 5NSeS eS[މS[މSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1kW{W{)M  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  YY)G*G44>1?1IIST^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeYY M M  AA55))  "#%%'')u*u,,.a/a113M4M668999;;=%>%@@BCEEGGIqJqLLNYOYQQSATAVVX)Y)[[]^``bbdmemggiUjUllnIoIqqs1t1vvxy{{}~uu]]чщEE-- b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYS)˃S)˃SS SSS:S:S~S~S S SSpSwSwSSS)QS`)QSeZSeZSS SG S| G S/S /SW~S W~S5NS 5NSkS kSaމSaމS S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1M} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1M} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1M k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1we5Y)  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10^^  LL)D*D44>0?0IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1^^LL : :  (({{i i""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIXJXLLNBOBQQS,T,VVXY[[]^`u`ubbd_e_ggiIjIlln?o?qqs)t)vvxy{{}}rr\\чщFF00 b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYSʃSʃS S  SSS:S:S~S~S[S[SSpSSSSSQS`QSZSZSS SG S| G SS S?~S ?~SNS NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Dx b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1"} {  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  SS)?*?44>%?%IIS T ^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33## { {""$k%k'')S*S,,.;/;113#4#668 9 ;;==?g@gBBDOEOGGI5J5LLNOQQSSUoVoXXZS[S]]_7`7bbdeggiikqlqnnpaqassuEvExxz){)}} ccՉՋGG++ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYS)˃S)˃SS SSS:S:SSS S SSpS}S}SSSqQS`qQSZSZSS SG S| G SS SE~S E~S#NS #NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1M} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1M} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1M k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1Hl:}Ko  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  KK)=*=44>%?%IIS T ^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__KK 7 7  ##qq] ]""$P%P''):*:,,.$/$11346688:m;m==?W@WBBDAEAGGI)J)LLNOQQSSUhVhXXZN[N]]_4`4bbdeggijlslsnnpeqessuKvKxxz1{1}}ppVVɎɐ<< b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations1SSuʃSuʃSS|SSSSS"S|"S&SH&S9S9SQ~SPQ~SSXSuSuS SP SSSeSeSS$ SoS, oSS SS$ SS SKDS| KDSURSXURS[S|[SٚSٚSSxSS\SSS%S%S/S8/SSSAlSAlSySySeSleS5S5SS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Py b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations15 bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Py b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations15  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Py  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations15 bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`*NEih;Bf  0RvZ~ $H$Hr"<` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Ei#G; -| Rvb$HHl.R`2V8(mm?c5z L p   B    Y }  + O   # h  :^ 0uGk=Tx&Ja3W)n@d6{MqC b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dE# rOAY_c(m2w<   F   P   Z d bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10^^ LL)D*D44>2?2IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1^^LL : : wwee{{i i""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIZJZLLNDODQQS.T.VVXY[[]^`w`wbbdaeaggiKjKllnAoAqqs+t+vvxy{{}}tt^^ӇӉHH22# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSA˃SA˃SSdSSlSi!Si!S$S$S:S@:SSSISISᐄS@ᐄS_S _S+S+SSSS SYS YSSd SCS CSqQS qQSZS ZSSS%DžS%DžSSSWSWSU'S U'SSSSpS/lSt/lSySTyS)S)S߉S$߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p< b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 E  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p< b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 E  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p<  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 E   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ko:@d)MBc+ vAek TxCIm b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@%IY}#UylOsFjIm DCgg=aAe7[?1U+O% I   C   =   7   1  +%~!|zxvtrpnlj h f db`^ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d%Yn47 fI?qJ#|W  2  h C   yT/  bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  PP)*44>'?'IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33##o o""$[%['')E*E,,.///1|1|335f6f88:P;P==?b@bBBDLELGGI7J7LLNOQQSTV{V{XXZc[c]]_K`Kbbd?e?ggi'j'llnoqqssukvkxxzS{S}};;##mm  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1SSuʃSuʃSS|SSSSS"S|"S&SH&S9S9SK~SPK~SSXSuSuS SP SSS_S_SS$ SoS, oSS SS$ SS SKDS| KDSwQSXwQSZS|ZSSSSxSS\SSS%S%SS8SSS)lS)lSySyS Sl S߉S߉SqS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ov b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations15 bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ov b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations15  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ov  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations15 bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` .k=am&Ae0Tc;%ISw Z~ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Ei; Rv$H_1U'l>bSw49]y  C K o   A  )  x    X | ^  * N D h   ! * N f  48\ .sGk=Tx&Ja3W)n@d6{MqC b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dE OY.c}(Lm2jw<   F   P   \ !f bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__ KK)=*=44>'?'IIS T ^^hhrrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__KK 7 7 pp\\qq] ]""$P%P''):*:,,.$/$11346688:m;m==?W@WBBDAEAGGI+J+LLNOQQSSUjVjXXZP[P]]_6`6bbdeggijlulunnpgqgssuMvMxxz3{3}}rrXXˎː>># b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSA˃SA˃SSdSSlSi!Si!S$S$S:S@:SSSISISᐄS@ᐄS_S _S1S1SSSS SYS YSSd SCS CSRS RS'\S '\SoSoSc˅Sc˅SSSSS(S (SSSSpS#lSt#lSySTyS S S߉S$߉S_S(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t@ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 E  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t@ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 E  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t@  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 E   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`z -Fjs<(LUyTx .7[{=].R b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@l @<`\| 0,PbLp<`  : l    ]  @ 7 [  < `  5 ~   \ X |  | 0,PLpj80TLph6.RJnf4,PH l  d!!!"""#2### b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dlXD0[$tm`6L8  "  v ` J 4r\ bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  LL)*44>$?$IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%qq[ [""$I%I'')3*3,,./1i1i335S6S88:=;===?P@PBBD:E:GGI$J$LLN O Q}Q}SSUcVcXXZI[I]]_/`/bbdeggijlulunnpYqYssu=v=xxz!{!}}ww]]ЉЋCC))  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSʃSʃSSdSSlSo!So!S$S$S:S@:S~S~SOSOS琄S@琄SeS eSSSSSS S_S _SSd S%CS %CSIRS IRS[S [S͚S͚S+ʅS+ʅSSSGSGSE(S E(SSSSpS/lSt/lSySTyS;S;S S$ SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 $E  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 $E  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 $E   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`)MQuf7[=azB*Sw(>wg.R b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ Df'xt% Vz78\%I?7[!nImPt[m   ,   >  , P  > b  Ptbt!7)M?cUyk 0"F8\Nrdz)?1 U  G!k!!!]"" b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d fnK;0Uz_D)sZ A (   v  ]D+y bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  RR)>*>44>?IIST^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbVV J J  >>22""  ""$p%p'')\*\,,.H/H113444668 9 ;;= > @@BBDhEhGGI(J(LLNOQQSSUVXY[|[|]]_d`dbbdVeVggi>j>lln&o&qqstvvxxzj{j}}RRƂƄ::!!zz  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSS5˃S5˃SSdSSlSo!So!S$S$S:S@:SSSOSOS琄S@琄SeS eS%S%SSSS S_S _SSd S%CS %CSYQS YQSZS ZSݙSݙSąSąSSSSS&S &SSSSpS/lSt/lSySTySGSGSS$SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e1 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 EW  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e1 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 EW  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e1  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 EW   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ei;Bf=@^X|9]d,P7>b b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ DPt"8\m"FKo .)M~+fv NrTx6Z2VB*z b  J n  2 V   >   & v  ^Jn6Z"F2p \Hl4X D 0nZ~Fj2VB b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d Pk 6p@Lb2R"tF   j  <   ` 2V bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  QQ)?*?44>/?/IIST^^hhrusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccQQ ? ?  --  n n""$\%\'')J*J,,.8/8113&4&6689;;=>@y@yBBDgEgGGIWJWLLNEOEQQS3T3VVX!Y![[]^``bbdtetggibjbllnMoMqqs1t1vvxy{_{_}}CC''33  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSʃSʃS Sd SSlSo!So!S$S$S:S@:S~S~SOSOS琄S@琄SeS eSSSSSS S_S _SSd S%CS %CSRS RS9\S 9\SSSɅSɅSSS5S5S3(S 3(SSSSpSlStlSySTyS S S߉S$߉S;S(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 #E  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 #E  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 #E   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` -|?cl@)2Vh\/S\ <` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@fz(<,P@dTxh2V8| *   g  > I m  . R + O  B f 1 ~   V z `   j~,@,P<`Lp\l|(8$H4XDhTxdt   !0!!!"@" b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dfL2~"doJ<0 V|` D (   p  T8d bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  NN)6*644>>HIST^z^zhhr^s^} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbRR B B  22""j j""$R%R''):*:,,."/"113 4 6~6~88:f;f==?L@LBBD0E0GGIIK]L]NNPAQASSUNVNXXZ2[2]]_`bbdegtgtiikXlXnnp>   b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSS;˃S;˃SSdSSlSo!So!S$S$S:S@:SSSOSOS琄S@琄SeS eS+S+SSSS S_S _SSd S%CS %CSQS QS[S [S[S[SSSSSSS%S %SSSSpS#lSt#lSySTyS S S߉S$߉SkS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i5 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 EV  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i5 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 EV  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i5  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 EV   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`W{2PtMq8\jCg DRv9],:^| b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Ei; Rv$H_1U'l4>b4y] K o ; _   A  =     X | f  * N D h   " F e  7[ -rDh: Qu#G^0T&k=a3xJn@ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dE OYcd(/mz2Ew<   F   P   Z d bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  KK)5*544>!?!IIS T ^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaKK 5 5    ~~hhR R""$<%<'')&*&,,./11335o6o88:Y;Y==?C@CBBD-E-GGIJLLNOQxQxSSUbVbXXZL[L]]_6`6bbd e ggi j llnnpgqgssuKvKxxz{}x}x\\΄ΆiiۉۋMM11  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLSSTS{!S{!S$S$S#:S(#:S~S~S[S[SS(SqSqSSSSSS SkS kSSL S1CS 1CSCRS CRS[S [SǚSǚS_S_S S S)SD)SSS;lS ;lSSSSS0SSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 E bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 E bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 E  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`(LEidCg8\t ?cHl 1;_Y}!E b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@7`>b@iGk%I bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1^)t? } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  TT)F*F44>8?8IIS$T$^^h i s}s}} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbTT F F  88**  "#%y%y'')k*k,,.]/]113O4O668A9A;;=3>3@@B%C%EEGHJJLMO~O~QQSlTlVVXZYZ[[]H^H``b4c4eeghjjlmo{o{qqsatavvxGyG{{}-~-kk݌ݎOO b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS/˃S/˃SSLSSTS!S!S$S$S:S(:S S SaSaSS(SwSwSSSSSS SqS qSSL S7CS 7CSQS QSZS ZS7S7SSSSS#SD#SSS5lS 5lSYSYS)S0)SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z& b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Ok bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z& b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Ok bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z&  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Ok  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ei<Fj5Y>Y}%I/S[9)1U b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@;jLp.R4cEi bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1d1~K } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  [[)G*G44>3?3IIST^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1gg[[ O O  CC77++  "#%%'')u*u,,.a/a113M4M668999;;=%>%@@BCEEGGIsJsLLN[O[QQSCTCVVX+Y+[[]^``bbdoeoggiWjWlln?o?qqs't'vvxy{{}}kk߂߄SSLJlj;;## b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLSSTS{!S{!S$S$S#:S(#:S~S~S[S[SS(SqSqSSSSSS SkS kSSL S1CS 1CSQS QS[S [SaSaSOSOSSSSDSSS)lS )lSSS߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 E bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 E bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fz  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 E  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`,Y}Aem /Qu7[pGk%:^f=a( b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@'rPt. R 0 {  Y } 7 [  9  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 U k6 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  PP)>*>44>,?,IIST^^hhrhsh} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$[%['')I*I,,.7/7113%4%6689;;=>@x@xBBDfEfGGITJTLLN>O>QQS(T(VVXY[[]]_q`qbbdWeWggi=j=lln!o!qqstvtvtxxzV{V}}88nnߎߐPP b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS5˃S5˃SSLSSTS!S!S$S$S:S(:SSSaSaSS(SwSwS%S%SSSS SqS qSSL S7CS 7CSQS QS+[S +[SsSsSSSSS SD SSSlS lS S S߉S0߉SYS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^* b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Pl bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^* b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Pl bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^*  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Pl  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ei7Z~;3Vz  /RvRv$H+Nr b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@'K -z \ > b D  & s  U y  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1e2L } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  UU)=*=44>%?%IIS T ^{^{hhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%% } }""$i%i'')Q*Q,,.9/9113!4!668 9 ;};}==?e@eBBDMEMGGI5J5LLNOQQSSUoVoXXZS[S]]_7`7bbdeggiikqlqnnpUqUssu9v9xxz{}}ssWWɉɋ;; b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTS!S!S %S %S):S():S~S~SmSmSS(SSSSSSSS S}S }SSL SCCS CCSRS RS\S \SWSWS;S;SSS/SD/SSSAlS AlSSSkS0kSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 G  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 G  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 G   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Gk(Le 6Zq)MBf}'iNr -3 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@*uOs)M'pJn$H"k bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1Qc, } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  RR)B*B44>2?2IIST^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbRR B B  22"" z z""$j%j'')Z*Z,,.J/J113:4:668*9*;;=>@@B C EEGGIrJrLLN^O^QQSJTJVVX6Y6[[]"^"``bceeggipjplln\o\qqsHtHvvx4y4{{} ~  nnZZБГ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S)˃S)˃SSLSSTS!S!S%S%S:S(:SSSySySS(SSSSS}S}SS SS SSL SOCS OCSQS QS1[S 1[SySySSSSSSDSSS#lS #lS S S߉S0߉S_S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ZL. bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ZL. bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}O  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ZL.  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`=yFj>kW{+O]X|#Os,P b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@2gMq3W=#r X| bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1],{J } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  OO)9*944>!?!IIST^z^zhhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((tt^ ^""$H%H'')2*2,,./11346{6{88:e;e==?M@MBBD3E3GGIJLLNNPrQrSSUXVXXXZ>[>]]_$`$bbd e g}g}iikclcnnpIqIssu/v/xxz{}}nnTTƉƋ88 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTS!S!S %S %S):S():S~S~SmSmSS(SSSSSSSS S}S }SSL SCCS CCSQS QSZS ZS=S=S S SSSSDSSS/lS /lS#S#S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 F bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 F bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 F  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ei+Cgn0T5<`x%I .Fjq:8 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@EiC f @ d  >  a ; _  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1u>P } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  NN):*:44>&?&IIST^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&tt` `""$L%L'')8*8,,.$/$11346688:r;r==?^@^BBDJEJGGI6J6LLNOQQSTVzVzXXZb[b]]_J`Jbbd2e2ggijllnoqvqvssu^v^xxzF{F}}..rrZZ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S/˃S/˃SSLSSTS!S!S%S%S:S(:SSSSSS(SSSSS}S}SS SS S SL  SUCS UCSRS RS[S [SSS͵S͵S'S'SSDSSSlS lS S SQ߉S0Q߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 [K0 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 [K0 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 [K0  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`<d:(Lt.R 8\Fj 0Hl@^ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@>b$H . }  c I m / S  9  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1h7U } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  MM)3*344>?IISS]j^jhhrLsL} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((ppV V""$<%<'')"*",,./1{1{335a6a88:G;G==?+@+BBD E G~G~IIK`L`NNPBQBSSU$V$XXZ[]w]w__aYbYddf;g;iiklnnpprpspuuwRxRzz|4}4hh؋؍HH b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTS!S!S/%S/%S/:S(/:S~S~SSS)S()SSSSS S S S  SS SSL SgCS gCSsRS sRS[S [SSSSS S S)SD)SSS;lS ;lS_S_S/S0/SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  ]@ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  ]@ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H|  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  ]@  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ei&J`.Rih6ZpC>bx Cg b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$rVz:^B&t X|<` bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1Nj8 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  PP)>*>44>,?,IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$[%['')I*I,,.7/7113%4%6689;;=>@x@xBBDfEfGGITJTLLN>O>QQS(T(VVXY[[]]_q`qbbd[e[ggiEjElln/o/qqstvvxy{x{x}}bbׂׄLL66  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S#˃S#˃SSLSSTS!S!S%S%S:S(:S~S~SsSsS S( SSSSSwSwSS SS SSL SICS ICSRS RS[S [S9S9SSSSS#SD#SSS5lS 5lS/S/S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1rD b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Z`" bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1rD b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Z`" bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1rD  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Z`"  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`>Ko%I)Mg+EiEiv %IeVz b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@)zdNr8\"F 0k bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1V'xI } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  TT)>*>44>)?)IIST^^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggWW G G  77''s s""$]%]'')G*G,,.1/111346689;z;z==?d@dBBDNENGGI9J9LLN!O!QQS T V}V}XXZe[e]]_M`Mbbd5e5ggijllnoqyqyssuavaxxzI{I}}11yycc b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTS!S!S/%S/%S):S():S~S~SSS)S()SSSSSSS S  SS SSL SgCS gCSQS QSZS ZSCSCSSSSSSDSSS)lS )lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  \> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  \> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H|  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  \>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ei /Jny 2V(LW{$H)4Xs:4Os b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ D(v Z ~ > b " F  * x  \  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1T"p> } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  OO)9*944>#?#IIS T ^|^|hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''uua a""$L%L'')6*6,,. / 113 4 6688:i;i==?S@SBBD=E=GGI'J'LLN O QQSSUfVfXXZL[L]]_2`2bbdeggiikqlqnnpWqWssu=v=xxz#{#}} ||bbՉՋHH.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S)˃S)˃SSLSSTS!S!S%S%S:S(:S~S~SsSsS S( SSSSSwSwSS SS SSL SICS ICSRS RS[S [SSSSS3S3SSDSSS)lS )lS S S߉S0߉S_S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1tH b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 [a$ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1tH b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 [a$ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1tH  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 [a$  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Azn$Ael $H^eRvHl?l b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Uy?c) M  7 ! r \ F j 0T bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1k< ^ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  PP):*:44>$?$IIS T ^{^{hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))uu_ _""$I%I'')3*3,,./11346|6|88:f;f==?P@PBBD:E:GGI$J$LLN O Q}Q}SSUcVcXXZI[I]]_/`/bbdeggiikilinnpMqMssu1v1xxz{}}kk݄݆QQĉċ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS!SL!SSTS!S!S$S$S5:S(5:S~S~SaSaSS(SwSwSSSSSS SqS qSSL S7CS 7CSRS RS[[S [[SSS+S+SCSCS)SD)SSS;lS ;lSGSGSS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 E  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 E  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I}  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 E   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`%Ip`Bf,Pe8 2V}k Qu6Zo b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$ ysmga[Uy bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1[4 f } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  VV)B*B44>,?,IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffZZ N N  BB66&&  ""$t%t'')`*`,,.L/L113848668$9$;;=>@@BBDlElGGITJTLLN3?3IIS!T!^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggUU C C  11  r r""$`%`'')N*N,,.@}@}BBDkEkGGI[J[LLNIOIQQS7T7VVX%Y%[[]^``bcexexggifjfllnOoOqqs3t3vvxy{{}}mm߂߄QQÇÉ55 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS!SL!SSTS!S!S$S$S/:S(/:S~S~SaSaSS(SwSwSSS S SS SqS qSSL S7CS 7CSQS QSZS ZS+S+SSSSSSDSSS#lS #lS S S߉S0߉SGS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  E  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  E  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I}  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  E   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`2mPt?c8\j5 DzRv,Nr:^| b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@OsI m C g = a 7 [ 1 U +O%I bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1?qJ# } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  RR):*:44> ? IIST^v^vhhrZsZ} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&&n n""$V%V'')>*>,,.&/&11346688:j;j==?P@PBBD4E4GGIJLLNNPnQnSSURVRXXZ6[6]]_`bbddfpgpiikTlTnnp8q8ssuvxxz{}r}rVVȄȆ:: b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS˃S˃SSLSSTS!S!S$S$S:S(:S~S~SaSaSS(SwSwS S SqSqSS SqS qSSL S7CS 7CSRS RS[S [S-S-SSSiSiSSDSSS/lS /lS S S߉S0߉SwS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i=  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 E P bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i=  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 E P bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i=   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 E P  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Bp+(Fj#,P{n&Txs *N| b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@+O! E  ; 1  ' ~  t j  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations13 a8 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  OO)9*944>%?%IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeOO 9 9  ##  llV V""$@%@'')***,,./11335s6s88:];]==?G@GBBD1E1GGIJLLNOQ|Q|SSUfVfXXZP[P]]_:`:bbd$e$ggijllnnpgqgssuKvKxxz/{/}}iiۉۋMM11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS'SL'SSTS!S!S%S%S5:S(5:S~S~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CSQS QS+[S +[SsSsSӯSӯSmSmSASDASSSSlS SlSSSkS0kSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 F  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 F  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 F   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`"FtX|>RvPt/S6Hl / .@dy b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$4 D0T@dPt` p bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1dH, } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  XX)J*J44>7@@B)C)EEGHJJL M OOQQSpTpVVX^Y^[[]L^L``b8c8eeg"h"jjl m ooqqsetevvxKyK{{}1~1ooSSőœ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS˃S˃S{SL{SSTS!S!S$S$S:S(:S~S~SaSaSS(SwSwSSSeSeSS SqS qSSL S7CS 7CS%RS %RSa[S a[SSSSSgSgS;SD;SSSMlS MlSqSqSAS0ASS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\. b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 F W bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\. b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 F W bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\.  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 F W  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`8?cw+)=ah%'.R}fu b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$ 0<$H0T<`HlTx bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1bD& } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  __)K*K44>7?7IIST^^his{s{} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kk__ S S  GG;;//##  "#%%'(*y*y,,.e/e113Q4Q668=9=;;=)>)@@BCEEGHJwJwLLN_O_QQSGTGVVX/Y/[[]^``bbdsesggi[j[llnCoCqqs+t+vvxy{{}}ooWWˇˉ??'' b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS!SL!SSTS!S!S%S%S5:S(5:S~S~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CSQS QS[S [SaSaSUSUSSS)SD)SSS;lS ;lS)S)S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 F bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 F bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 F  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`5mVz)?cJn5>bzsLp2Vg b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ ,  < ( L 8 \ H l X|hx bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1tX<  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  TT)B*B44>0?0IIST^^hhrlsl} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffTT B B  00  q q""$_%_'')M*M,,.;/;113)4)6689;;=>@|@|BBDjEjGGIXJXLLNBOBQQS,T,VVXY[[]^`u`ubbd[e[ggiAjAlln%o%qqstvxvxxxzZ{Z}}<<rrTT b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS˃S˃S{SL{SSTS!S!S$S$S:S(:S~S~SaSaSS(SwSwSSSkSkSS SqS qSSL S7CS 7CS=RS =RSy[S y[SSSSSSS#SD#SSS5lS 5lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^2 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 E W bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^2 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 E W bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^2  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 E W  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`;b* #Gn/Suz$i;_ 0 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@` l  x " .  : "F.R:^ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1jL.r } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  YY)A*A44>)?)IIS T ^^hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiYY I I  99))   ""$m%m'')U*U,,.=/=113%4%668 9 ;;==?i@iBBDQEQGGI9J9LLNOQQSTVsVsXXZW[W]]_;`;bbdeggijlulunnpYqYssu=v=xxz!{!}}ww[[͉͋??## b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS-SL-SSTS!S!S %S %S;:S(;:S~S~SmSmSS(SSSSSSSS S}S }SSL SCCS CCSQS QSZS ZS S S{S{SSSGSDGSSSYlS YlSSS}S0}S%S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 E bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 E bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 E  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`9m>b?c Dh5&zJnsLp-QVz b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$&JLpr)+OQuw . bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1o^M< } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  VV)F*F44>6?6IIS"T"^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&& ~ ~""$n%n'')^*^,,.N/N113>4>668.9.;;=>@@BCEEGGIvJvLLNbObQQSNTNVVX:Y:[[]&^&``bceeggitjtlln`o`qqsLtLvvx8y8{{}$~$rr^^ԑԓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS ˃S ˃SoSLoSSTS!S!S %S %S:S(:S~S~SmSmSS(SSSSS_S_SS S}S }SSL SCCS CCS_QS _QSZS ZS㙅S㙅S]S]SSS#SD#SSS5lS 5lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q# b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 IZ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q# b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 IZ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q#  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 IZ  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ei --Qj=,PPtTx6Oss&J b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$"FDhf=;_] bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1mZG4 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  SS)=*=44>%?%IIS T ^~^~hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,xxb b""$L%L'')6*6,,. / 113 4 6688:i;i==?Q@QBBD7E7GGIJLLNOQvQvSSU\V\XXZB[B]]_(`(bbdeggiikglgnnpMqMssu3v3xxz{}}rrXXʉʋ<<  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS'SL'SSTS!S!S %S %S;:S(;:S~S~SmSmSS(SSSSSSSS S}S }SSL SCCS CCSQS QSZS ZS S SSSMSMS/SD/SSSAlS AlS5S5SS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 E bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 E bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 E  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` 0nJny @d(LVz6+5Y{v Mq7Tx b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@0 T V z |  3 5 Y [8:^ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1+ xg } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  RR)>*>44>*?*IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **xxd d""$P%P'')<*<,,.(/(11346689;v;v==?b@bBBDNENGGI:J:LLN"O"QQS T V~V~XXZf[f]]_N`Nbbd6e6ggijllnoqzqzssubvbxxzJ{J}}22vv^^ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS˃S˃SuSLuSSTS!S!S%S%S:S(:S~S~SgSgSS(S}S}SSSeSeSS SwS wSSL S=CS =CSRS RS=[S =[SSSoSoS;S;SSDSSS#lS #lS S Sc߉S0c߉S S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S' b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 H [ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S' b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 H [ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S'  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 H [  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ei4Tx=+Rv)TxPt&J'Nr b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ 4 2 V T x v  + )MKom" bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1!{hU } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  QQ)7*744>?IISS]n^nhhrPsP} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,ttZ Z""$@%@'')&*&,,. / 11335e6e88:K;K==?/@/BBDEGGIIKdLdNNPFQFSSU(V(XXZ [ ]{]{__a]b]ddf?g?iik!l!nnpqststuuwVxVzz|8}8ll܋܍LL b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃSWSWSS$Se:Se:S~S~SaS$aSݩSݩSASASS$SQSQSZS4ZSST Se S e SSS SS+S +SelS elSS SYSYSS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1%Y] b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1%Y] b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1%Y] b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 kAe;X|  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ww  ee)S*S44>???IIS)T)^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1wwee S S  AA//   ""$p%p'')^*^,,.L/L113:4:668(9(;;=>@@BCE{E{GGIgJgLLNQOQQQS;T;VVX%Y%[[]^``bbdnenggiXjXllnBoBqqs,t,vvxy{{}~uu__ԇԉII33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS ˃S ˃SoS oSSS}:S}:S~S~SISISSpSYSYSSSeQS`eQSZSZSS$ Se S e SAS ASS SSlS SlSMS MSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1i} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1i} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1i k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ei; Rv$H  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  aa)K*K44>4?4IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ttdd T T  DD44!!   ""$j%j'')T*T,,.>/>113(4(6689;;==?q@qBBD[E[GGIDJDLLN,O,QQSTVVXXZp[p]]_X`Xbbd@e@ggi(j(llnoqqssulvlxxzT{T}}<<$$ ll b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃSWSWSS$Sk:Sk:S~S~SaS$aS㩄S㩄SGSGSS$SQSQSZS4ZSST Se S e S;S ;SS SMlS MlSS S߉S߉SS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1&Z[ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1&Z[ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1&Z[ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 i;_ 1vHl  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10vv  bb)L*L44>4?4IIST^^hissss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvbb N N  ::&&t t""$_%_'')I*I,,.3/311346689;|;|==?f@fBBDPEPGGI8J8LLNOQQSTVwVwXXZ][]]]_C`Cbbd)e)ggijllnnphqhssuNvNxxz4{4}}ssYY̎̐?? b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS ˃S ˃SoS oSSS:S:S~S~SISISSpS_S_SSSeQS`eQSZSZSS$ Se S e S5S 5S S  SGlS GlS S  S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1h} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1h} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ey b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1h k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ei; Rv$H  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  __)I*I44>1?1IIST^^hhrlsl} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1tt`` L L  88$$n n""$X%X'')B*B,,.,/,11346689;u;u==?_@_BBDIEIGGI1J1LLNOQQSSUpVpXXZV[V]]_<`v>xxz"{"}}xx\\ωϋBB(( b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃS?S ?SSSM:SM:S~S~SaSaSũSpũS)S)SSSQS`QS [S [SS$ S_ S _ SAS ASS SSlS SlS_S _S/S/SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ei4X&J<  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ww  gg)S*S44>;?;IIS#T#^^h i s}s}} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1wwkk _ _  SSGG77##  "#%%'')q*q,,.]/]113I4I668595;;=>@@BCE{E{GGIcJcLLNKOKQQS3T3VVXY[[]^`w`wbbd_e_ggiGjGlln/o/qqstvvxxzs{s}}[[ςτCC(( b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS˃S˃SuS uSSS:S:S~S~S7S7SSpS_S_SSSeQS`eQSZSZSS$ S_ S _ SAS ASS SSlS SlSqS qSASASS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Z} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Z} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Z k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ei; Rv$H  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  ^^)L*L44>:?:IIS(T(^^h i s~s~} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp^^ L L  ::(( { {""$i%i'')W*W,,.E/E113343668!9!;;=>@@BBDtEtGGIbJbLLNPOPQQS>T>VVX,Y,[[]^``bceeggimjmllnVoVqqs:t:vvxy{{}~ttXXʇʉ<<  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃS9S 9SSSM:SM:S~S~S[S[SũSpũS)S)SSS[RS`[RS[S[SS$ S_ S _ S)S )SS S;lS ;lS S  S߉S߉S_S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ei] u%>  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10uu  aa)I*I44>-?-IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1uuee U U  EE55!!  } }""$e%e'')M*M,,.5/511346689;y;y==?]@]BBDAEAGGI%J%LLN O Q{Q{SSU_V_XXZC[C]]_'`'bbd e g}g}iikalannpEqEssu)v)xxz { }}ccՄՆEE''  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS˃S˃SuS uSSS:S:S~S~S7S7SSpSeSeSSSkQS`kQSZSZSS$ S_ S _ S5S 5S S  SGlS GlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N" b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1[} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N" b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1[} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N" b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1[ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ei=W{+O  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  ZZ)D*D44>.?.IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ppZZ D D  ..wwa a""$K%K'')5*5,,./113 4 6~6~88:h;h==?R@RBBDG?GIIS3T3^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ssee W W  II;;--  "#%%'(*|*|,,.n/n113`4`668R9R;;=D>D@@B6C6EEG(H(JJLMOOQRT{T{VVXiYi[[]W^W``bCcCeeg-h-jjlmooqqsptpvvxVyV{{}<~<""xx\\ΑΓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS˃S˃SuS uSSS:S:S~S~SCSCSSpSeSeSSS_QS`_QSZSZSS$ S_ S _ SAS ASS SSlS SlSwS wSGSGSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W) b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1k} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W) b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1k} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W) b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1k k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ei; Rv$H  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  dd)P*P44>:?:IIS"T"^^h i s~s~} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ppdd X X  LL@@44((  "#%%'(*~*~,,.j/j113V4V668B9B;;=.>.@@BCEEGHJzJzLLNbObQQSJTJVVX2Y2[[]^``bcevevggi^j^llnFoFqqs.t.vvxy{{}}rrZZ·ΉBB** b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS3S 3SSSG:SG:S~S~SaSaSSpS#S#SSSRS`RS[S[SS$ S_ S _ S5S 5S S  SGlS GlS5S 5SSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 EiiCCg  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  __)M*M44>9?9IIS!T!^^hissss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qq__ M M  ;;)) | |""$j%j'')X*X,,.F/F113444668"9";;=>@@BBDuEuGGIaJaLLNKOKQQS5T5VVXY[[] ^ `~`~bbddedggiJjJlln.o.qqstvvxxzc{c}}EE''yy[[ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS˃S˃S{S {SSS:S:S~S~SCSCSSpSeSeSSS_QS`_QSZSZSS$ S_ S _ S)S )SS S;lS ;lS S  S߉S߉SwS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W+ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1j} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W+ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1j} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W+ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1j k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ei; Rv$H  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  ``)H*H44>.?.IIST^^hhrhsh} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp`` P P  @@00    "#%t%t'')\*\,,.D/D113,4,6689;;==?p@pBBDXEXGGI>J>LLN"O"QQSTVxVxXXZ\[\]]_@`@bbd$e$ggijlzlznnp^q^ssuBvBxxz&{&}} ||``҉ҋDD(( b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS-S -SSSA:SA:S~S~SmSmSSpSSSSSQS`QSZSZSS$ SY S Y SGS GSS SYlS YlSS S}S}S%S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ei@2$y  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10oo  __)O*O44>=?=IIS)T)^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1oo__ O O  ??//  ""$w%w'')g*g,,.W/W113G4G668797;;='>'@@BCEEGHJ}J}LLNiOiQQSUTUVVXAYA[[]-^-``bceeghj{j{llngogqqsStSvvx?y?{{}+~+yyeeۑۓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS˃S˃S{S {SSS:S:S~S~SSSSpSkSkSSSRS`RS=[S=[SS$ SY S Y SS SS S/lS /lS S  S߉S߉SkS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`2 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1M} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`2 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1M} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`2 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1M k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 yk]Os  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  TT)>*>44>$?$IIS T ^}^}hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiUU A A  --yyc c""$M%M'')7*7,,.!/!113 4 6688:j;j==?P@PBBD6E6GGIJLLNOQuQuSSU[V[XXZA[A]]_'`'bbd e ggiikflfnnpLqLssu2v2xxz{}}qqUUljNj99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS-S -SSS;:S;:S~S~SmSmSSpSSSSSRS`RS[S[SS$ SY S Y S/S /SS SAlS AlS5S 5SSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Eim%)MQu  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  YY)E*E44>/?/IIST^^hhrsss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mmYY E E  11  k k""$W%W'')C*C,,.///11346689;};}==?i@iBBDUEUGGI?J?LLN'O'QQSTVVXXZk[k]]_S`Sbbd;e;ggi#j#lln o qqssugvgxxzO{O}}77{{cc b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS˃S˃SS SSS:S:S~S~SSSSpSkSkSSSRS`RS=[S=[SS$ SY S Y S S  SS SlS lS S  S]߉S]߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`4 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1N} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`4 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1N} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`4 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1N k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 |n `Rv  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  TT):*:44>?IISS]o^ohhrOsO} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kkWW C C  //ww] ]""$C%C''))*),,./11335h6h88:N;N==?0@0BBDEGGIIKeLeNNPGQGSSU)V)XXZ [ ]|]|__a^b^ddf@g@iik"l"nnpqsusuuuwWxWzz|9}9kkۋۍKK b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS'S 'SSS;:S;:S~S~SaSaSSpSSSSSQS`QSZSZSS$ SS S S S5S 5S S  SGlS GlSkS kS;S;SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ImC . u  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  YY)G*G44>3?3IIST^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kkYY G G  55##v v""$d%d'')R*R,,.@/@113.4.6689;;= > @@BBDoEoGGI[J[LLNEOEQQS/T/VVXY[[]^`x`xbbdbebggiLjLlln6o6qqs t vvx y {{}}iiނބSSȇȉ=='' b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS˃S˃SS SSS:S:S~S~S7S7S Sp SqSqSSSQS`QS7[S7[SS$ SS S S S#S #SS S5lS 5lS/S /S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i; b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1h} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i; b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1h} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i; b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1h k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 yl ^Pt  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  UU)?*?44>(?(IIST^^hhrlsl} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhXX H H  88((t t""$^%^'')H*H,,.2/211346689;{;{==?e@eBBDOEOGGI8J8LLN O QQSTV|V|XXZd[d]]_L`Lbbd4e4ggijllnoqxqxssu`v`xxzH{H}}00vv`` b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS'S 'SSS5:S5:S~S~SaSaSSpSSSSSsRS`sRS[S[SS$ SS S S S#S #SS S5lS 5lSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 EiiCCg  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  VV)@*@44>(?(IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjVV B B  ..||h h""$S%S'')=*=,,.'/'11346688:p;p==?Z@ZBBDDEDGGI,J,LLNOQQSSUkVkXXZQ[Q]]_7`7bbdeggijlvlvnnp\q\ssuBvBxxz({(}}ggډڋMM33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS˃S˃SS SSS:S:S~S~S7S7S Sp SqSqSSS RS` RSI[SI[SS$ SS S S SS SS S)lS )lS S  S߉S߉SeS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i=  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1g} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i=  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1g} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i=  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1g k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 &{m _  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  SS)=*=44>%?%IIS T ^|^|hhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,xxb b""$L%L'')6*6,,. / 113 4 6688:i;i==?S@SBBD=E=GGI%J%LLN O Q~Q~SSUdVdXXZJ[J]]_0`0bbdeggiikjljnnpNqNssu2v2xxz{}}llބކPPÉË66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃS!S !SSS5:S5:S~S~SaSaSSpSSSSSRS`RS[[S[[SS$ SS S S S)S )SS S;lS ;lSAS ASSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ~ zvt  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  [[)G*G44>/?/IIST^^hhrqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kk__ S S  GG;;++  "#%y%y'')e*e,,.Q/Q113=4=668)9);;=>@@BBDoEoGGIWJWLLN?O?QQS'T'VVXY[[]]_k`kbbdSeSggi;j;lln#o#qqs t vvxxzg{g}}OOÂÄ77uu b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S#˃S#˃SS SSS:S:S~S~SSSSpSwSwSSSQS`QS1[S1[SS$ SS S S S)S )SS S;lS ;lSSS SS#S#SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1rD b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1K} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1rD b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1K} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1rD b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1K k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 yk]Os  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  RR)@*@44>.?.IIST^^hisrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddRR @ @  ..  o o""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGIVJVLLNDODQQS2T2VVX Y [[]^``bbdsesggiajallnJoJqqs.t.vvxy{{}}hhڂڄLL00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃSS SSS/:S/:S~S~SaSaSSpS S SSSwQS`wQSZSZSS$ SS S S SS SS S#lS #lS S  S߉S߉SAS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 i;_ 1vHl  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  UU)=*=44>!?!IIST^w^whhrYsY} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiYY I I  99))q q""$Y%Y'')A*A,,.)/)11346688:m;m==?Q@QBBD5E5GGIJLLNNPoQoSSUSVSXXZ7[7]]_`bbddfqgqiikUlUnnp9q9ssuvxxz{}s}sWWɄɆ99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S#˃S#˃SS SSS:S:S~S~SSSSpSwSwSSSRS`RS[S[SS$ SS S S SS SS S)lS )lS S  S߉S߉SqS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1rF b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1K} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1rF b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1K} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1rF b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1K k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 "F>bZ~ v  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  NN)8*844>"?"IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddNN 8 8  ""  kkU U""$?%?''))*),,./11335r6r88:\;\==?F@FBBD0E0GGIJLLNOQyQySSUcVcXXZM[M]]_7`7bbd!e!ggi j llnnpdqdssuHvHxxz,{,}}ff؉؋JJ.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃSS SSS/:S/:S~S~SUSUSSpS S SSSQS`QS1[S1[SS$ SM S M S/S /SS SAlS AlSS S_S_SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 yk]Os  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  YY)K*K44>;?;IIS'T'^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggYY K K  ==//!!  "#%~%~'')p*p,,.b/b113T4T668F9F;;=8>8@@B*C*EEGHJJL M OOQQSoToVVX]Y][[]K^K``b7c7eeg!h!jjl m ooqqsftfvvxLyL{{}2~2nnRRđē b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S)˃S)˃SS SSS:S:S~S~S+S+SSpSwSwSSSQS`QSZSZSS SM S| M S/S /SW~S W~S5NS 5NSeS eS[މS[މSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{M b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1e} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{M b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1e} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{M b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1e k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1|,se  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  XX)D*D44>.?.IIST^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddXX L L  @@44((  "#%%'')r*r,,.^/^113J4J668696;;=">"@@BCEEGGInJnLLNVOVQQS>T>VVX&Y&[[]^``bbdjejggiRjRllnDoDqqs,t,vvxy{{}}ppXẊ̉@@(( b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17L_SʃSʃSS SSS):S):S~S~SUSUSSpSSSSSwQS`wQSZSZSS$ SM S M SS SS S/lS /lSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 k=a3xJn  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>-?-IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIUJULLN?O?QQS)T)VVXY[[]]_r`rbbdXeXggi>j>lln&o&qqstvyvyxxz[{[}}==qqSS b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S)˃S)˃SS SSS:S:SSS+S+SSpS}S}SSSRS`RS[S[SS SM S| M SS S?~S ?~SNS NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1d} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1d} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{O b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1d k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1ImAe   b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  TT)<*<44>"?"IIST^^hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$ | |""$h%h'')P*P,,.8/8113 4 6689;|;|==?d@dBBDLELGGI2J2LLNOQQSSUlVlXXZP[P]]_4`4bbdeggiiknlnnnpZqZssu>v>xxz"{"}}xx\\Ή΋@@$$ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYSʃSʃSS SSS#:S#:S~S~SUSUSSpSSSSSwQS`wQSZSZSS SG S| G S;S ;Sc~S c~SANS ANSS SމSމS?S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1t nRv  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  SS)E*E44>3?3IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33## { {""$m%m'')]*],,.M/M113=4=668-9-;;=>@@B C EEGGIsJsLLN_O_QQSKTKVVX7Y7[[]#^#``bceeggiqjqllnioiqqsUtUvvxAyA{{}-~-{{ggݑݓ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYS/˃S/˃SS SSS:S:SSS S SSpS}S}SSSRS`RS[S[SS SG S| G SS S?~S ?~SNS NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V" b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1K} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V" b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1K} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V" b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1K k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1&J 1U{  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10]]  RR)<*<44>"?"IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1]]II 5 5  !!  wwa a""$K%K'')5*5,,./113 4 6~6~88:h;h==?N@NBBD4E4GGIJLLNOQsQsSSUYVYXXZ?[?]]_%`%bbdeggiikplpnnpVqVssu'?'IIST^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%ss_ _""$O%O'');*;,,.'/'11346688:u;u==?a@aBBDMEMGGI7J7LLNOQQSTV{V{XXZc[c]]_K`Kbbd3e3ggijllnoqqssukvkxxzS{S}};;## gg b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYS/˃S/˃SS SSS:S:S S S S SSpSSSSSwQS`wQSZSZSS SG S| G SS S-~S -~S NS  NS S  S}݉S}݉S%S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X$ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1K} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X$ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1K} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X$ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1K k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1Z~VzRvNr  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  PP)6*644>?IIST^w^whhrWsW} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__KK 7 7  ##ssY Y""$?%?'')%*%,,. / 1~1~335d6d88:J;J==?,@,BBDEGGIIKaLaNNPCQCSSU%V%XXZ[]x]x__afbfddfHgHiik*l*nnp q s}s}uuw_x_zz|A}A##ssSSÐÒ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations1SS{ʃS{ʃSS|SSSSS"S|"S&SH&S9S9SQ~SPQ~SSXSuSuS SP SSSeSeSS$ SoS, oSS SS$ SS SKDS| KDSRSXRS[S|[S S Si…Sxi…S%S\%SSS%S%S/S8/SSSAlSAlSySySeSleS5S5SS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P{ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations17 bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P{ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations17  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P{  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations17 bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`An)$Ae 1$He9HlA?l b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@GkA^4X .uKo!Eb8\ 2 y   O s  % I    f  < `  6} Swg)MMq&3Wm=Cg=$s Z~Y}0T?c*qGkA^4X b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dGUc*q8F 7T    V b % ) t  p  7 ~ G bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa UU)K*K44>7?7IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaUU I I %%  " # %%'(*y*y,,.e/e113Q4Q668=9=;;=)>)@@BCEEGHJwJwLLN_O_QQSGTGVVX/Y/[[]^``bbdsesggi[j[llnOoOqqs7t7vvxy{{}~{{ccׇ׉KK33# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSA˃SA˃SSdSSlSo!So!S$S$S:S@:SSSOSOS琄S@琄SeS eS1S1SSSS S_S _SSd S%CS %CSSQS SQSZS ZSיSיSgąSgąS#S#SySySw&S w&S5S5S Sp SGlStGlSySTySSS_S$_SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t@ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 G bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t@ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 G bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t@  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 G  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Gk&6Z_A44X}2`W{{ 6Z1U b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Nr2V:lPt4X< nRv6 Z   >   " p    4 T x m  8 \ ' K  @>b(xW{5`nHl(L0T<$t \Dh,P8 pX|@d(L b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dNj8T"p>N \@ , y  | 2  L   l <  \, bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  VV) * 44>7@@B)C)EEGHJJL M OOQQSpTpVVX^Y^[[]L^L``b8c8eeg"h"jjl m ooqqsqtqvvxWyW{{}=~=## {{__ёѓ  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1SSuʃSuʃSS|SSSSS"S|"S&SH&S9S9SQ~SPQ~SSXSuSuS SP SSSeSeSS$ SoS, oSS SS$ SS SKDS| KDSQSXQS%[S|%[SmSmS1ąSx1ąSS\SOSOSM&SM&SS8SSS)lS)lSySyS Sl S߉S߉SeS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ox b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations17 bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ox b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations17  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Ox  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations17 bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`#GvZ~CbVz?NrRv:^=Pt b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Nr2V:lPt4X< nRv6 Z   >   " p   T x  8 \  @ lLp.R2~^>bB"%In /Pt~0Td4Jn0T`@d: D o b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dNj8T"p> X$  p  <   T C    l a80 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa QQ)A*A44>)?)IIS T ^^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaQQ A A ~~nn y y""$m%m'')U*U,,.=/=113%4%668 9 ;;==?i@iBBDQEQGGI9J9LLNOQQSTVsVsXXZW[W]]_;`;bbdeggijlulunnpeqessuIvIxxz-{-}}ggىًKK//# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSG˃SG˃SSdSSlSo!So!S$S$S:S@:S#S#SOSOS琄S@琄SeS eS7S7SSSS S_S _SSd S%CS %CSRS RS[S [SSSȅSȅSaSaSSS'S 'SSSSpS/lSt/lSySTySSS߉S$߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xD b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 G bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xD b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 G bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xD  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 G  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`q Dh(8\*N 0PtPt$Hhv<` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@a ku(2<"F,P6 Z  @ d  J n  T x  ^  hr|#)Cg /!Z~58s;/SA Fj#G&)M/S5Y;_AeGkMqSwY }  ! b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1daB#eF'iJ)   g k  F $  % ]    OcB! bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  PP)*44>0?0IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$_%_'')M*M,,.;/;11335u6u88:c;c=>@|@|BBDjEjGGIXJXLLNBOBQQS,T,VVXY[[]^`u`ubbd[e[ggiAjAlln1o1qqstvvxxzf{f}}HH** ~~``  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSʃSʃSSdSSlSu!Su!S$S$S#:S@#:S~S~SUSUS퐄S@퐄SkS kSSSSSS SeS eSSd S+CS +CSRS RS[S [S?S?S̅S̅SSS%S%S#)S #)SSSSpS#lSt#lSySTyS S S߉S$߉SYS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  G bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  G bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  G  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` D\t 1X|Qu Tx=aPt=m)Im b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@p(,PTx |48\` @  D h  l  $  ( L  Pt@d"Fx 0(u4X W{\;_BB$qJnv 08\d&JRv ~8@dl !&!!!."R"""Z#~##$$$%@% b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dp`P@0 p!` n  R ;  D  U  6  (   ~pbT bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10^^  II)3*344>>HIS T ^~^~hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1^^JJ 6 6  ""nnX X""$B%B''),*,,,./11346u6u88:_;_==?G@GBBD-E-GGIIK\L\NNPBQBSSURVRXXZ8[8]]_`bbdeggiikglgnnpMqMssu3v3xxz{}}rrXXʉʋ<<   b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYS;˃S;˃SSdSSlSu!Su!S$S$S:S@:SSSUSUS퐄S@퐄SkS kS+S+SSSS SeS eSSd S+CS +CSRS RSC[S C[SSSi…Si…S%S%SSS%S %S/S/SSpSAlStAlSySTySSSeS$eS S(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i5 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 GY bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i5 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 GY bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i5  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 GY  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Gk)>bg@OsEiW{#Gi%_Gk b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@GkA^4X .uKo!Eb8\ 2 y   O s  % I     f N r  < ` 0 k  6)M}  SwFj)M(c#!Ej@d:W{-Q'nDh>[1U b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dGUc*q8F] T U    b K )  p  7 ~ E  bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  TT)D*D44>4?4IIS T ^^hhrs} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$ | |""$l%l'')\*\,,.L/L113<4<668,9,;;=>@@B C EEGGItJtLLN`O`QQSLTLVVX8Y8[[]$^$``bceeggirjrlln`o`qqsLtLvvx8y8{{}}pp\\rr^^ԑԓ  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSʃSʃSSdSSlSu!Su!S$S$S#:S@#:S~S~SUSUS퐄S@퐄SkS kSSSSSS SeS eSSd S+CS +CSWSS WSS\S \SۛSۛS̅S̅ScScSSS)S )SSSSpS lSt lSySTyS{ S{ SK߉S$K߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 G bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 G bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 G  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`"FX|2V>Pt8Jn3W 0Bf4X b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@q*0TZ~=Cgm&, P  V z     9  ? c  i"(LRv |35Y[Sw5Y8;:^j`Lp.R=4?ceBD h  j!!!!"""##G###I$m$$% b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dqbSD5&yj  Y  H  7 "  & o   < sVbQ@ bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  II)/*/44>>HISS]n^nhhrPsP} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``LL 8 8  $$llR R""$8%8'')*,,./1w1w335]6]88:C;C==?'@'BBD E GzGzIIK1L1NNPQSSU V XXZ[]s]s__a]b]ddf?g?iik!l!nnpqststuuwVxVzz|8}8ll܋܍LL  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSA˃SA˃SSdSSlSu!Su!S$S$S:S@:SSSUSUS퐄S@퐄SkS kS1S1SSSS SeS eSSd S+CS +CSRS RS[S [SSSi…Si…S%S%SSS%S %SSSSpS)lSt)lSySTySSS߉S$߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m9 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 GX bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m9 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 GX bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m9  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 GX  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Lp%In$,PIm 0$H|m#\Hl#Gl b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@GkA^4X .uKo!Eb8\ 2 y   O s  % I    f  < `  6} Sw)M#j@d>b: [=W{x-Q6Z'nSwDh5p>[1U b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dGUc*q8F T   b )   p A |  7 7  rF  bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  NN):*:44>&?&IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&tt` `""$L%L'')8*8,,.$/$11346688:r;r==?^@^BBDJEJGGI6J6LLNOQQSTVzVzXXZb[b]]_J`Jbbd2e2ggijllnoqzqzssubvbxxz{}}{{vv^^  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLSSTS{!S{!S$S$S):S():S~S~S[S[SS(SqSqSSSSSS SkS kSSL S1CS 1CSRS RS-\S -\SuSuS?S?SSSSDSSS)lS )lS#S#S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 G bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 G bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H|  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 G  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`#GGkz<0T_!&JTx@9].Ru b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ `>b@iGk%I'r bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1J  ` +  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  NN)8*844>#?#IIS T ^^hhrisi} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaQQ A A  11!!m m""$W%W'')A*A,,.+/+11346688:t;t==?^@^BBDHEHGGI3J3LLNOQQSTVwVwXXZ_[_]]_G`Gbbd/e/ggijllnnpsqsssu[v[xxzC{C}}++ss]] b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS5˃S5˃SSLSSTS!S!S$S$S:S(:SSSaSaSS(SwSwS%S%SSSS SqS qSSL S7CS 7CSQS QS[S [SISISSS-S-S#SD#SSS5lS 5lS_S_S/S0/SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^* b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Me bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^* b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Me bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^*  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Me  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Gk .Hlv <`*NX|0T 0:^}$H@_ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@& J I m /l.RQu7t bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1g$a  [  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  RR)@*@44>.?.IIST^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddRR @ @  ..  o o""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGIVJVLLN@O@QQS*T*VVXY[[]]_s`sbbd]e]ggiGjGlln1o1qqstvvxy{z{z}}ddقلNNÇÉ88"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLSSTS{!S{!S$S$S):S():S~S~S[S[SS(SqSqSSSSSS SkS kSSL S1CS 1CS7RS 7RSs[S s[SSS/ÅS/ÅSSS SD SSSlS lS S S߉S0߉SYS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 G bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 G bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H|  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 G  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`!EdVz9pJn -|?c&"3Wx2 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Pt.R 0{Y}7[9b bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 v A W" } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  LL)6*644> ? IIST^w^whhr]s]} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%qq[ [""$E%E'')/*/,,./11346x6x88:b;b==?L@LBBD6E6GGI J LLNOQyQySSU_V_XXZE[E]]_+`+bbdeggiikelennpIqIssu-v-xxz{}}ggلنMM33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS;˃S;˃SSLSSTS!S!S$S$S:S(:SSSaSaSS(SwSwS%S%SSSS SqS qSSL S7CS 7CSRS RS[S [SESESSSSSSDSSS#lS #lS S S߉S0߉SkS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b. b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Nf bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b. b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Nf bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b.  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Nf  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Gk*NtA 0Z~<``6Z"Fl(Rv b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@6ZY}?|>b$a#G  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 U  O  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  OO)9*944>#?#IIS T ^|^|hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''uua a""$L%L'')6*6,,. / 113 4 6688:i;i==?S@SBBD=E=GGI'J'LLN O QQSSUfVfXXZL[L]]_2`2bbdeggiikqlqnnpWqWssu=v=xxz#{#}} ||bbՉՋHH.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTS!S!S %S %S/:S(/:S~S~SmSmSS(SSSSS S SS S}S }SSL SCCS CCSRS RS9\S 9\SSSٻSٻSSS#SD#SSS5lS 5lSMSMSS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 I  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 I  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I}  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 I   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Gk&Jz<b$9]Tx<`+O.Rx b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ 7Z~4X2{ Uy/S - bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1`) r ;   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  OO)=*=44>-?-IIST^^hhrqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaOO = =  ++~~l l""$Z%Z'')H*H,,.6/6113$4$6689;;=>@w@wBBDeEeGGIUJULLNCOCQQS1T1VVXY[[] ^ ``bbdrerggi`j`llnIoIqqs-t-vvxy{{}}ggقلKK// b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S/˃S/˃SSLSSTS!S!S%S%S:S(:S S SySySS(SSSSSSSS SS SSL SOCS OCS RS  RSI[S I[SSSSSSS#SD#SSS5lS 5lSASASS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 XN. bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 XN. bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 XN.  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`3Rv$ ?c{r +Os^;gJn b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ D Uy#GX|&J[)M bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations12u8 { >  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  XX)D*D44>.?.IIST^^hhrrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hh\\ P P  DD88((  "#%v%v'')b*b,,.N/N113:4:668&9&;;=>@@BBDnEnGGIVJVLLN>O>QQS&T&VVXY[[]]_j`jbbdReRggi:j:lln"o"qqs t v~v~xxzf{f}}NN‚„66vv b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃSSLSSTS!S!S %S %S):S():S~S~SmSmSS(SSSSSSSS S}S }SSL SCCS CCSQS QS[S [SISIS…S…SSSSDSSS#lS #lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 H bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 H bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I}  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 H  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` DGkTx*N4Dh D%7[{:)Mm b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@vPt*N(qKo%I#l bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 M  _ (q } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  KK)5*544>!?!IIS T ^}^}hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaKK 5 5    ~~hhR R""$<%<'')&*&,,./11335o6o88:Y;Y==?C@CBBD-E-GGIJLLNOQxQxSSUbVbXXZL[L]]_6`6bbd e ggi j llnnpcqcssuGvGxxz+{+}}ee׉׋II-- b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S5˃S5˃SSLSSTS!S!S%S%S:S(:S S SSSS(SSSSSSSS SS S SL  SUCS UCSRS RS-\S -\SuSuSgSgS_S_S SD SSSlS lS S S߉S0߉SAS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W# b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 YM0 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W# b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 YM0 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W#  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 YM0  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`/Sz&<k& -Tx&Ei&.R b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@^,Pa/S!d2V bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1  D  J  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  RR):*:44> ? IIST^v^vhhrZsZ} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&&n n""$V%V'')>*>,,.&/&11346688:j;j==?P@PBBD4E4GGIJLLNNPnQnSSURVRXXZ6[6]]_`bbddfpgpiikTlTnnp8q8ssuvxxz{}r}rVVȄȆ:: b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS!SL!SSTS!S!S/%S/%S5:S(5:S~S~SSS)S()SSSSSSS S  SS SSL SgCS gCS SS  SSE\S E\SSSἅSἅS#S#S)SD)SSS;lS ;lSeSeS5S05SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  _@ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  _@ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  _@  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Gk)M|@wi3 -\ ?cIm <`o b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@d(L4lTx<`$H 0 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 \ , | L  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  YY)E*E44>1?1IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeYY M M  AA55))  "#%%'')s*s,,._/_113K4K668797;;=#>#@@BCEEGGIqJqLLNYOYQQSATAVVX)Y)[[]^``bbdmemggiUjUlln=o=qqs%t%vvx y {{}}ii݂݄QQŇʼn99!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS)˃S)˃SSLSSTS!S!S%S%S:S(:S~S~SsSsS S( SSSSSwSwSS SS SSL SICS ICSRS RS!\S !\SiSiSSS/S/S;SD;SSSMlS MlSSSeS0eS S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1tH b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 XZ" bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1tH b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 XZ" bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1tH  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 XZ"  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`8\0T8\8\\8\B b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@fJn.R6hLp0T bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 n < X  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  ZZ)L*L44>>?>IIS*T*^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhZZ L L  >>00""  "#%%'')q*q,,.c/c113U4U668G9G;;=9>9@@B+C+EEGHJJL M OOQQSrTrVVX`Y`[[]N^N``b:c:eeg$h$jjlmooqqsgtgvvxMyM{{}3~3qqUUǑǓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS!SL!SSTS!S!S/%S/%S/:S(/:S~S~SSS)S()SSSSS S S S  SS SSL SgCS gCSQS QS [S  [SUSUSąSąS7S7SSDSSS)lS )lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  ^> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  ^> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  ^>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`GkCRv /<`>b2V*Nm(L:Y} b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@hPt8\ D,|dLp bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 l < \ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  UU)=*=44>%?%IIS T ^{^{hhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%% } }""$i%i'')Q*Q,,.9/9113!4!668 9 ;};}==?e@eBBDMEMGGI5J5LLNOQQSSUoVoXXZS[S]]_7`7bbdeggiikqlqnnpUqUssu9v9xxz{}}ssWWɉɋ;; b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS)˃S)˃SSLSSTS!S!S%S%S:S(:SSSsSsS S( SSSSS}S}SS SS SSL SICS ICS'SS 'SSc\S c\SSS%ąS%ąS7S7S#SD#SSS5lS 5lS#S#S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vL b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Y[$ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vL b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Y[$ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vL  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Y[$  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`4X)$H|  Dwl4mh  0X|  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@8jNr2V:lPt bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1& t B^ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  TT)B*B44>0?0IIST^^hhrlsl} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffTT B B  00  q q""$_%_'')M*M,,.;/;113)4)6689;;=>@|@|BBDjEjGGIXJXLLNBOBQQS,T,VVXY[[]^`u`ubbd[e[ggiAjAlln%o%qqstvxvxxxzZ{Z}}<<rrTT b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS'SL'SSTS!S!S$S$S5:S(5:S~S~SaSaSS(SwSwSSSSSS SqS qSSL S7CS 7CSSS SSW\S W\SSS뾅S뾅SSSSDSSS)lS )lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  G bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  G bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  G  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`,Pk~ DAep;6d*Y}VzJn b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@!ECA?=;97 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1|W 2 h  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  MM)7*744>?IIST^x^xhhr^s^} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&rr\ \""$F%F'')0*0,,./11346y6y88:c;c==?K@KBBD1E1GGIJLLNNPpQpSSUVVVXXZ<[<]]_"`"bbdeg{g{iikalannpGqGssu-v-xxz{}}ll߄߆RRĉċ66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS#˃S#˃SSLSSTS!S!S$S$S:S(:S~S~SaSaSS(SwSwS S SqSqSS SqS qSSL S7CS 7CSRS RS[S [S S S S SSSASDASSSSlS SlSSSwS0wSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i=  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 GM bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i=  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 GM bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i=   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 GM  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`=f?-Qz(+OAe<<`Uy,PPt b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@]W{QuKoEi?c9]3W bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 g @  r  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  XX)H*H44>8?8IIS$T$^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhXX H H  88((  ""$p%p'')`*`,,.P/P113@4@668090;;= > @@BCEEGHJxJxLLNdOdQQSPTPVVX?IISS]j^jhhrLsL} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((ppV V""$<%<'')"*",,./1{1{335a6a88:G;G==?+@+BBD E G~G~IIK`L`NNPBQBSSU$V$XXZ[]w]w__aYbYddf;g;iiklnnpprpspuuwRxRzz|4}4hh؋؍HH b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS#˃S#˃SSLSSTS!S!S$S$S:S(:S~S~SaSaSS(SwSwSSSwSwSS SqS qSSL S7CS 7CS]SS ]SS\S \SᛅSᛅSƅSƅS S S)SD)SSS;lS ;lS/S/S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kA  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 GN bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kA  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 GN bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kA   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 GN  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`9] 00T,7[).R%5Y,P b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@-Q'K!E?93 -' bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 K $}V/ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  RR)>*>44>*?*IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **xxd d""$P%P'')<*<,,.(/(11346689;v;v==?b@bBBDNENGGI:J:LLN"O"QQS T V~V~XXZf[f]]_N`Nbbd6e6ggijllnoqzqzssubvbxxzJ{J}}22vv^^ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS-SL-SSTS!S!S%S%S;:S(;:S~S~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CS-SS -SSi\S i\SSSSSaSaS)SD)SSS;lS ;lS5S5SS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 H bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 H bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 H  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`/Sy$Haw?Iml41Uc+X|  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$8(L<`Ptdx&: bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1h N 4   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  RR)<*<44>'?'IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%%q q""$[%['')E*E,,.///11346689;x;x==?b@bBBDLELGGI7J7LLNOQQSTV{V{XXZc[c]]_K`Kbbd3e3ggijllnoqwqwssu_v_xxzG{G}}//wwaa b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS˃S˃S{SL{SSTS!S!S%S%S:S(:S~S~SgSgSS(S}S}SSSkSkSS SwS wSSL S=CS =CSmRS mRS[S [SSSSSmSmS;SD;SSSMlS MlSqSqSAS0ASS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^2 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 GQ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^2 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 GQ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^2  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 GQ  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`>h 02V, 0Jn($Hc&=a b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@dPt` p,<(L8\ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1| ` D (  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  VV)D*D44>2?2IIST^^his{s{} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhVV D D  22  s s""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIZJZLLNDODQQS.T.VVXY[[]^`w`wbbdaeaggiKjKlln5o5qqstvvx y {~{~}}hh݂݄RRLJlj<<&& b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS'SL'SSTS!S!S%S%S;:S(;:S~S~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CSQS QSZS ZS%S%SɅSɅSSSSDSSS/lS /lS S S߉S0߉SkS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 H bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 H bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 H  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`ArNr%Im2Vs  D:W{a=ap b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@*N>bRvfz( < ,!P!!! bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1f L2~ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  PP):*:44>$?$IIS T ^{^{hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))uu_ _""$I%I'')3*3,,./11346|6|88:f;f==?P@PBBD:E:GGI$J$LLN O Q}Q}SSUcVcXXZI[I]]_/`/bbdeggiikilinnpMqMssu1v1xxz{}}kk݄݆QQĉċ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS˃S˃SSLSSTS!S!S%S%S:S(:S~S~SgSgSS(S}S}S S SqSqSS SwS wSSL S=CS =CSRS RS-\S -\SuSuSɅSɅSSS#SD#SSS5lS 5lSSS߉S0߉S}S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`6 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 GQ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`6 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 GQ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`6  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 GQ  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`2Vx -8j%*Lp} >bu Dv b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@HlX|hx$4 D 0!T!!! bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1p T8 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  SS)=*=44>'?'IIS T ^^hhrfsf} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++yye e""$P%P''):*:,,.$/$11346688:m;m==?W@WBBDAEAGGI+J+LLNOQQSSUjVjXXZP[P]]_6`6bbdeggijlulunnp[q[ssuAvAxxz'{'}} ffىًLL22 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃS-SL-SSTS!S!S %S %SA:S(A:S~S~SmSmSS(SSSSSSSS S}S }SSL SCCS CCSRS RS!\S !\SiSiS…S…SSS5SD5S S SGlS GlSeSeS5S05SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 H bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 H bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 H  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`(Lrv4Hl\BDh*N_y7_ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@`CIms,2V\? bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1V G 8 )   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>1?1IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIYJYLLNGOGQQS5T5VVX#Y#[[]^``bbdvevggidjdllnMoMqqs1t1vvxy{{}}kk݂݄OO33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqS˃S˃SuSLuSSTS!S!S %S %S:S(:S~S~SmSmSS(SSSSSeSeSS S}S }SSL SCCS CCSRS RS=[S =[SSSÅSÅS S S;SD;SSSMlS MlSSSSS#S0#SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S' b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 GX bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S' b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 GX bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S'  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 GX  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Gk7Z~A3Vz /^Rv4X+Nr b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@3WY}68\^;=a bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1j Y H 7 &  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  \\)H*H44>2?2IIST^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ll`` T T  HH<<,,  "#%z%z'')f*f,,.R/R113>4>668*9*;;=>@@BBDrErGGIZJZLLNBOBQQS*T*VVXY[[]]_n`nbbdVeVggi>j>lln&o&qqstvvxxzj{j}}RRƂƄ::!!zz b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃS-SL-SSTS!S!S %S %SA:S(A:S~S~SmSmSS(SSSSSSSS S}S }SSL SCCS CCSQS QSZS ZSSSM̅SM̅SSS)SD)SSS;lS ;lS S S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 H bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 H bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 H  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`>sHlIm$HcC#>bl<`Fj b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Eio(.RX | !!!";"""A#e###k$$ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 |m^O } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  OO)9*944>%?%IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeOO 9 9  ##  llV V""$@%@'')***,,./11335s6s88:];]==?G@GBBD1E1GGIJLLNOQ|Q|SSUfVfXXZP[P]]_:`:bbd$e$ggijllnnpgqgssuKvKxxz/{/}}iiۉۋMM11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqS˃S˃S{SL{SSTS!S!S %S %S:S(:S~S~SmSmSS(SSSSSkSkSS S}S }SSL SCCS CCSCRS CRS[S [SǚSǚSM̅SM̅SSS#SD#SSS5lS 5lS S S߉S0߉SSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U+ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 GY bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U+ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 GY bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U+  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 GY  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Gk'Kb %I3Wn%?coz$KoKo 0 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@c@Bfh !!!!"E"""G#k###m$$ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1sbQ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  VV)>*>44>$?$IIST^z^zhhr^s^} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjZZ J J  ::**r r""$Z%Z'')B*B,,.*/*11346688:n;n==?T@TBBD8E8GGIJLLNOQrQrSSUVVVXXZ:[:]]_`bbdegtgtiikXlXnnp>  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃSQSQSS$S_:S_:S~S~SaS$aSשSשS;S;SS$SQSQSZS4ZSST Se S e SSS SS+S +SelS elSS SYSYSS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1!U_ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1!U_ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1!U_ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 kAe;X|  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10zz  nn)Z*Z44>D?DIIS,T,^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1zznn b b  VVJJ>>22  "&#&%%'(**,,.t/t113`4`668L9L;;=8>8@@B$C$EEGHJJLLNlOlQQSTTTVVXI?IIIS5T5^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1uugg Y Y  KK==//!!  "#%%'(*~*~,,.p/p113b4b668T9T;;=F>F@@B8C8EEG*H*JJLMOOQRT}T}VVXkYk[[]Y^Y``bEcEeeg/h/jjlmooqqsrtrvvxXyX{{}>~>$$zz^^БГ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃSQSQSS$Se:Se:S~S~SaS$aSݩSݩSASASS$SQSQSZS4ZSST Se S e S;S ;SS SMlS MlSS S߉S߉SS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1"V] b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1"V] b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1"V] b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 mCg?\  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10xx  hh)P*P44>6?6IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1xxhh X X  HH88((  "#%|%|'')d*d,,.L/L1134446689;;=>@x@xBBD`E`GGIFJFLLN*O*QQSTVVXXZd[d]]_H`Hbbd,e,ggijllnnpfqfssuJvJxxz.{.}}hhډڋLL00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS˃S˃SuS uSSS:S:S~S~SISISSpSeSeSSSwQS`wQSZSZSS$ Se S e SAS ASS SSlS SlSAS ASSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1f} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1f} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1G{ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1f k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 GkA^4X  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10uu  cc)Q*Q44>=?=IIS%T%^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1uucc Q Q  ??--   ""$n%n'')\*\,,.J/J113848668&9&;;=>@@BCEyEyGGIeJeLLNOOOQQS9T9VVX#Y#[[] ^ ``bbdhehggiNjNlln2o2qqstvvxxzg{g}}II++ }}__ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃSES ESSSS:SS:S~S~S[S[S˩Sp˩S/S/SSSQS`QS%[S%[SS$ Se S e S/S /SS SAlS AlSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Gk?c7[/S  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  ^^)H*H44>.?.IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ss__ K K  77##m m""$W%W'')A*A,,.+/+11346688:t;t==?Z@ZBBD@E@GGI&J&LLN O QQSSUeVeXXZK[K]]_1`1bbdeggiikplpnnpVqVssu???IIS+T+^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qqaa Q Q  AA11!!  "#%y%y'')i*i,,.Y/Y113I4I668999;;=)>)@@BCEEG H JJLLNkOkQQSWTWVVXCYC[[]/^/``bceeghj}j}llnioiqqsUtUvvxAyA{{}-~-{{ggݑݓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃS?S ?SSSS:SS:S~S~S[S[S˩Sp˩S/S/SSSQS`QS7[S7[SS$ Se S e SS SS S/lS /lS S  So߉So߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 GkCg?c;_  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  \\)B*B44>$?$IIST^w^whhrWsW} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ss__ K K  77##  e e""$K%K'')1*1,,./11335p6p88:V;V==?8@8BBDEGGIIKmLmNNPOQOSSU1V1XXZ[]]__afbfddfHgHiik*l*nnp q s}s}uuw_x_zz|A}A##ssSSÐÒ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS˃S˃S{S {SSS:S:S~S~S7S7SSpSeSeSSSqQS`qQSZSZSS$ Se S e S;S ;SS SMlS MlSGS GSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P& b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Y} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P& b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Y} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P& b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Y k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 GkA^4X  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  ]])I*I44>3?3IIST^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qq]] I I  55!!  o o""$[%['')G*G,,.3/31134668 9 ;;==?m@mBBDYEYGGICJCLLN+O+QQSTVVXXZo[o]]_W`Wbbd?e?ggi'j'llnoqqssukvkxxzS{S}};;## gg b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃS9S 9SSSM:SM:S~S~SaSaSũSpũS)S)SSSQS`QS[S[SS$ S_ S _ S;S ;SS SMlS MlSAS ASSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Gk8\.R$H  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10rr  __)I*I44>2?2IIST^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1rrbb R R  BB22  ~ ~""$h%h'')R*R,,.e>ggi&j&llnoqqssujvjxxzR{R}}::"" jj b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS˃S˃S{S {SSS:S:S~S~S=S=SSpSkSkSSSqQS`qQSZSZSS$ S_ S _ SAS ASS SSlS SlSwS wSGSGSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y- b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1e} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y- b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1e} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y- b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1e k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 GkA_5Y  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  [[)I*I44>5?5IIST^^h i s~s~} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mm[[ I I  77%% x x""$f%f'')T*T,,.B/B1130406689;;= > @@BBDqEqGGI]J]LLNGOGQQS1T1VVXY[[]^`z`zbbddedggiNjNlln8o8qqs"t"vvx y {{}}kkUUʇʉ??)) b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃS9S 9SSSG:SG:S~S~SaSaSSpS#S#SSSRS`RS[S[SS$ S_ S _ S)S )SS S;lS ;lS S  S߉S߉SwS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Gko'+OSw  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  [[)E*E44>-?-IIST^^hhrhsh} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp\\ H H  44    j j""$T%T'')>*>,,.(/(11346688:q;q==?[@[BBDEEEGGI-J-LLNOQQSSUlVlXXZR[R]]_8`8bbdeggijlrlrnnpVqVssu:v:xxz{}}ttXXˉˋ>>$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS˃S˃SS SSS:S:S~S~S=S=SSpSkSkSSSqQS`qQSZSZSS$ S_ S _ S/S /SS SAlS AlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y/ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1d} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y/ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1d} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y/ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1d k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 GkA^4X  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  ZZ)D*D44>,?,IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nnZZ F F  22  l l""$W%W'')A*A,,.+/+11346688:t;t==?^@^BBDHEHGGI0J0LLNOQQSSUoVoXXZU[U]]_;`;bbd!e!ggijlzlznnp`q`ssuFvFxxz,{,}}kkމދQQĎĐ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS3S 3SSSG:SG:S~S~SmSmSSpS#S#SSSQS`QS[S[SS$ SY S Y S;S ;SS SMlS MlSeS eS5S5SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Gk*N D:  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  \\)J*J44>8?8IIS&T&^^h i s|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nn\\ J J  88&& y y""$g%g'')U*U,,.C/C1131416689;;= > @@BBDrErGGI`J`LLNNONQQS1?1IIST^^hissss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mmaa U U  II==--  "#%{%{'')g*g,,.S/S113?4?668+9+;;=>@@BBDqEqGGIYJYLLNAOAQQS)T)VVXY[[]]_m`mbbdUeUggi=j=lln%o%qqs t vvxxzi{i}}QQłń99ww b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS3S 3SSSA:SA:S~S~SmSmSSpSSSSSRS`RS\S\SS$ SY S Y S)S )SS S;lS ;lS S  S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Gk !E]  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  VV)@*@44>*?*IIST^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1llVV @ @  **ss] ]""$G%G'')1*1,,./11346z6z88:d;d==?N@NBBD8E8GGI"J"LLN O QQSSUkVkXXZU[U]]_?`?bbd)e)ggijllnnplqlssuPvPxxz4{4}}nnRRĎĐ66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS˃S˃SS SSS:S:S~S~S%S%S Sp SqSqSSSRS`RS[[S[[SS$ SY S Y SS SS S/lS /lS S  S߉S߉SMS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b8 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1P} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b8 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1P} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b8 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1P k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 (uk   b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  YY)A*A44>%?%IIS T ^{^{hhr]s]} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mm]] M M  ==-- u u""$]%]'')E*E,,.-/-11346688:q;q==?U@UBBD9E9GGIJLLNOQsQsSSUWVWXXZ;[;]]_`bbdeguguiikYlYnnp=q=ssu!v!xxz{}w}w[[̈́͆== b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS-S -SSSA:SA:S~S~SaSaSSpSSSSSQS`QSZSZSS$ SY S Y S5S 5S S  SGlS GlSkS kS;S;SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Gk D; 1  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  bb)N*N44>8?8IIS T ^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nnbb V V  JJ>>22&&  "#%%'(*|*|,,.h/h113T4T668@9@;;=,>,@@BCEEGHJxJxLLN`O`QQSHTHVVX0Y0[[]^``bcetetggi\j\llnDoDqqs,t,vvxy{{}}ppXẊ̉@@(( b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS#˃S#˃SS SSS:S:S~S~S1S1SSpSwSwSSSRS`RSO[SO[SS$ SY S Y S;S ;SS SMlS MlSS SeSeS S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1k?  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1b} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1k?  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1b} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1k?  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1b k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 vl b  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  [[)M*M44>=?=IIS)T)^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ii[[ M M  ??11##  "#%%'')r*r,,.d/d113V4V668H9H;;=:>:@@B,C,EEGHJJL M OOQQSqTqVVX_Y_[[]M^M``b9c9eeg#h#jjl m ooqqsftfvvxLyL{{}2~2nnRRđē b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS'S 'SSS;:S;:S~S~SaSaSSpSSSSSRS`RS[S[SS$ SY S Y SS SS S/lS /lS S  S߉S߉SkS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Gk@Tx  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  \\)D*D44>*?*IIST^^hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ll\\ L L  <<,,   ""$p%p'')X*X,,.@/@113(4(6689;;==?l@lBBDTETGGI:J:LLNOQQSTVtVtXXZX[X]]_<`v>xxz"{"}}xx\\Ή΋@@$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS#˃S#˃SS SSS:S:S~S~S1S1SSpSwSwSSSRS`RS[S[SS$ SY S Y S#S #SS S5lS 5lS#S #S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kA  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1a} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kA  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1a} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kA  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1a k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 &JBf^z  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  WW)E*E44>1?1IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiWW E E  33!!t t""$b%b'')P*P,,.>/>113,4,6689;;=>@@BBDmEmGGIYJYLLNCOCQQS-T-VVXY[[]^`v`vbbd\e\ggiBjBlln&o&qqstvyvyxxz[{[}}==qqSS b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS'S 'SSS;:S;:S~S~SaSaSSpSSSSSgRS`gRS[S[SS$ SS S S SS SS S)lS )lS S  S߉S߉S_S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 4B,P:^  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  RR)<*<44>"?"IIST^{^{hhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++wwa a""$K%K'')5*5,,./113 4 6~6~88:h;h==?N@NBBD4E4GGIJLLNOQsQsSSUYVYXXZ?[?]]_%`%bbd e g~g~iikdldnnpJqJssu0v0xxz{}}ooSSʼnŋ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS)˃S)˃SS SSS:S:S~S~SSSSpSwSwSSSRS`RSO[SO[SS$ SS S S S5S 5S S  SGlS GlSS SqSqSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1tH b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1L} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1tH b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1L} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1tH b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1L k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 #zpf  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  UU)E*E44>3?3IIST^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%% } }""$m%m'')]*],,.M/M113=4=668-9-;;=>@@B C EEGGIsJsLLN_O_QQSKTKVVX7Y7[[]#^#``bceeggiqjqlln]o]qqsItIvvx5y5{{}!~! oo[[ёѓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS!S !SSS5:S5:S~S~SaSaSSpSSSSSQS`QS[S[SS$ SS S S SS SS SlS lS S  SW߉SW߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 rVz:^C  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  PP)6*644>?IISS]k^khhrKsK} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++ssY Y""$?%?'')%*%,,. / 1~1~335d6d88:J;J==?,@,BBDEGGIIKaLaNNPCQCSSU%V%XXZ[]x]x__aZbZddf'?'IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))wwc c""$O%O'');*;,,.'/'11346688:u;u==?a@aBBDMEMGGI7J7LLNOQQSTV{V{XXZc[c]]_K`Kbbd3e3ggijllnoqwqwssu_v_xxzG{G}}//ss[[ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃS!S !SSS/:S/:S~S~SUSUSSpS S SSS=RS`=RSy[Sy[SS$ SM S M SS SS S/lS /lS)S )S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 (, 04  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  SS)=*=44>&?&IIST^^hhrlsl} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&&r r""$\%\'')F*F,,.0/011346689;y;y==?c@cBBDMEMGGI6J6LLNOQQSTVzVzXXZb[b]]_J`Jbbd4e4ggijllnoqxqxssu`v`xxzH{H}}00vv`` b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S/˃S/˃SS SSS:S:SSS%S%SSpS}S}SSSQS`QSZSZSS SM S| M S/S /SW~S W~S5NS 5NSeS eS[މS[މSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}Q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1_} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}Q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1_} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}Q b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1_ k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1} .${q  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  OO)=*=44>)?)IIST^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaOO = =  ++~~l l""$Z%Z'')H*H,,.6/6113$4$6689;;=>@w@wBBDeEeGGIQJQLLN;O;QQS%T%VVXY[[]]_n`nbbdXeXggiBjBlln6o6qqs t vvx y {{}}iiނބSSȇȉ=='' b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17L_SʃSʃSS SSS/:S/:S~S~SUSUSSpS S SSSQS`QSZSZSS$ SM S M SS SS S#lS #lS S  S߉S߉S_S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 mCg=Z~  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  OO)9*944>!?!IIS T ^|^|hhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((tt^ ^""$H%H'')2*2,,./11346{6{88:e;e==?O@OBBD9E9GGI!J!LLNOQzQzSSU`V`XXZF[F]]_,`,bbdeggiikjljnnpNqNssu2v2xxz{}}llބކPPÉË66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S/˃S/˃SS SSS:S:S S S%S%SSpSSSSSSS`SSW\SW\SS SM S| M SS S?~S ?~SNS NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}S b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1^} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}S b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1^} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}S b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1^ k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1`:8\  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  NN)8*844> ? IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&tt` `""$K%K'')5*5,,./113 4 6~6~88:h;h==?R@RBBD.?.IIST^^h i s~s~} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGIVJVLLNDODQQS2T2VVX Y [[]^``bbdsesggiajallnVoVqqs:t:vvxy{{}~ttXXʇʉ<<  b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYS/˃S/˃SS SSS:S:S S S S SSpSSSSSaRS`aRS[S[SS SM S| M S)S )SQ~S Q~S/NS /NSGS GS=މS=މSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z& b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1I} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z& b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1I} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z& b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1I k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1x8h(<`  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  QQ)G*G44>/?/IIS#T#^^h i s}s}} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaUU I I  ==11!!   ""$y%y'')e*e,,.Q/Q113=4=668)9);;=>@@BBDoEoGGIWJWLLN?O?QQS'T'VVXY[[]]_k`kbbd_e_ggiGjGlln/o/qqstvvxxzs{s}}[[ςτCC(( b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYSʃSʃSS SSS):S):S~S~SUSUSSpSSSSSQS`QSZSZSS SM S| M SS SE~S E~S#NS #NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1H| b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1l1U+'K  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  JJ)8*844>"?"IIS T ^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``JJ 4 4  }}ggQ Q""$?%?''))*),,./11335r6r88:\;\==?F@FBBD0E0GGIJLLNOQyQySSUcVcXXZM[M]]_7`7bbd!e!ggi j llnnppqpssuTvTxxz8{8}}rrVVȎȐ:: b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYS5˃S5˃SS SSS:S:S S S S SSpSSSSSeQS`eQSZSZSS SM S| M SS S9~S 9~SNS NS S  S݉S݉SmS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\( b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1J} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\( b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1J} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\( b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1J k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1Qu;_Hl>b  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  MM)=*=44>!?!IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaQQ A A  11!!  i i""$Y%Y'')A*A,,.)/)11346688:m;m==?Q@QBBD5E5GGIJLLNNPoQoSSUSVSXXZ7[7]]_`bbd e g}g}iikalannpEqEssu)v)xxz { }}ccՄՆEE''  b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations1SS{ʃS{ʃSS|SSSSS"S|"S&SH&S9S9SW~SPW~SSXS{S{SSPSSSkSkSS$ SuS, uSS SS$ S S  SQDS| QDSQSXQSZS|ZS=S=S…Sx…SS\SSS%S%S;S8;SSSMlSMlSySySSlSqSqSS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P} b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations19 bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P} b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations19  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P}  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations19 bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`%3Wz\"^#GcDh-QRv(L b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@3Imr#G8\!"jaDh'KBPt:ey?c>a; _   9    \  6 Z   4 }  W{1U /x Rv,P*sMq'K%nHl"F iCgA b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dIN [L$ m6HZ#l  5 ~ G   Y  "k4 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__ OO)I*I44>9?9IIS%T%^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__OO ? ? {{kkw w""$q%q'')a*a,,.Q/Q113A4A668191;;=!>!@@BCEEGHJyJyLLNeOeQQSQTQVVX=Y=[[])^)``bceeghjwjwllnoooqqs[t[vvxGyG{{}3~3 mm# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSG˃SG˃SSdSSlSu!Su!S$S$S:S@:S#S#SUSUS퐄S@퐄SkS kS7S7SSSS SeS eSSd S+CS +CS RS  RSI[S I[SSS…S…SSSSS&S &S)S)SSpS;lSt;lSySTySSSSS#S$#SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xD b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 I  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xD b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 I  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xD  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 I   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Im;^#G@d" "FljFj(Lp b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@5Imn%I(L#l?cFjVz D4m'KgAe?b< `   :    ]  7 [   5 ~  X|2V 0y Sw-Q+tNr(L&oIm#G!jDhB b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dIJ]<&uo8J\%n  7  I   [  $m6 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  PP)*44>0?0IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$]%]'')K*K,,.9/911335r6r88:`;`=>@z@zBBDhEhGGIXJXLLNFOFQQS4T4VVX"Y"[[]^``bbdueuggicjcllnXoXqqs>""  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1SS{ʃS{ʃSS|SSSSS"S|"S&SH&S9S9SQ~SPQ~SSXS{S{SSPSSSeSeSS$ SuS, uSS SS$ S S  SQDS| QDSRSXRS[S|[SSSƅSxƅScS\cS-S-S+'S+'S#S8#SSS5lS5lSySyS)Sl)S߉S߉SS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Oz b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations19 bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Oz b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations19  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Oz  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations19 bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Bp&(Fj(,Pn*&Tx, *N| b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Y}SwMqGkAe;_5Y?c/S)h)M.R# G   Y }  A  C   ; H l   5  2   /  )#|rh^TxJn@d6Z,P"F<2(uk a b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dY2 d=J oHHF!zQ  (  V -   [2 ` bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__ KK)?*?44>+?+IIST^^his{s{} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__KK 7 7 llXXqq] ]""$Q%Q'')=*=,,.)/)11346689;w;w==?c@cBBDOEOGGI;J;LLN#O#QQS T VVXXZg[g]]_O`Obbd7e7ggijllnoqqssuovoxxzW{W}}??''kk# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSM˃SM˃SSdSSlSu!Su!S$S$S:S@:S)S)SUSUS퐄S@퐄SkS kS=S=SSSS SeS eSSd S+CS +CS=RS =RSy[S y[SSSŅSŅSSSSS&S &SSSSpS)lSt)lSySTyS S S߉S$߉SqS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|H b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 I bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|H b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 I bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|H  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 I  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`f7:^v Dl"*Rvy2:^aB b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@VzJn>b2V&J>2>b&|Uy3pl d & J    X | = a   L p T x  @ d  4 X  (L@ 0 tdTxDh4X$H8(|l\Lp<`,P@ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dV,X./hZ!Z0\0   X ,   T  (|P$ bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  JJ) * 44>$?$IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``JJ 4 4  }}ggQ Q""$?%?''))*),,./1[1[335E6E88:/;/==?F@FBBD0E0GGIJLLNOQ{Q{SSUeVeXXZO[O]]_9`9bbd#e#ggi j llnoqrqrssuVvVxxz:{:}}ttXXʎʐ<<  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSʃSʃSSdSSlS{!S{!S$S$S):S@):S~S~S[S[SS@SqS qSSSSSS SkS kSSd S1CS 1CSRS RS[S [SESES…S…SSSSS&S &S)S)SSpS;lSt;lSySTyS_S_S/S$/SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 I bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 I bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I}  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 I  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$5YY}C')MjHl8\\*r* b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Im2o#G1U!jTxDh:wB9]f\@d>a; _   :    ]  7 [   5 ~  X|2V 0y Sw-Q+tPt*N(qKo%I#lFj D b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dIK[E$m6HZ#l  5 ~ G   Y  "k4 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  MM);*;44>)?)IrIrST^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__MM ; ;  ))||j j""$X%X'')F*F,,.4/4113"4"6689;;==?u@uBBDcEcGGIQJQLLN;O;QQS%T%VnVnXXZX[X]]_B`BbbdXeXggiBjBlln6o6qqs t vvx y {{}}iiނބSSȇȉ==''  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSA˃SA˃SSdSSlS{!S{!S$S$S:S@:SSS[S[SS@SqS qS1S1SSSS SkS kSSd S1CS 1CS=RS =RSy[S y[SSS…S…SSSSS&S &S#S#SSpS5lSt5lSySTySYSYS)S$)SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m9 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 I[ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m9 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 I[ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m9  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 I[  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ko@d,P_t  <`s! 0Lp5@ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Im 0k#G)M! jGk)DhdB"Fe?c?c=`: ^   8    [   5 Y   3 |  Vz0T .wQu+O)rLp&J$mGk!E iCgA b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dI G[=$xm6HZ#l  5 ~ G   Y  "k4 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  ZZ)F*F44>2?2IIST^^hhrxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffZZ N N  BB66**  "#%%'')t*t,,.`/`113L4L668898;;=$>$@@BCEEGGIrJrLLNZOZQQSBTBVVX*Y*[[]^``bbdnenggiVjVlln@o@qqs(t(vvxy{X{X}}@@((ȇȉ<<$$  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSʃSʃSSdSSlS{!S{!S$S$S):S@):S~S~S[S[SS@SqS qSSSSSS SkS kSSd S1CS 1CSRS RS[S [S!S!S…S…SSSSS&S &SSSSpS#lSt#lSySTyS S S߉S$߉SkS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 I bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 I bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I}  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 I  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`!Emp*GkKo!E.Rs+hVz/S~ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Im#G!jDhBe?c D=Cg)`f: ^ ( L   8   K o   1 [  n  5 Y   3 |  Vz0T .wQu+O)rLp&J$mGk!EhBf@ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dI[$m?6|9Hv3Z#l  5 ~ G   [  $m6 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  LL)6*644> ? IfIfST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``LL 8 8  $$rr^ ^""$I%I'')3*3,,./11346|6|88:f;f==?P@PBBD:E:GGI$J$LLN O Q}Q}SSU6V6XXZ[]]_`bbdeggiiknlnnnp\q\ssuBvBxxz({(}}ggډڋMM33  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSG˃SG˃SSdSSlS{!S{!S$S$S:S@:SSS[S[SS@SqS qS1S1SSSS SkS kSSd S1CS 1CSwQS wQSZS ZSSS…S…SSSSS&S &S S SSpSlStlSySTyS S S߉S$߉S_S(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q= b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 IZ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q= b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 IZ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q=  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 IZ  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Im8<`#GBg!&jKosDh-QUy b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Im#G!jFj Dg!Ae\?>yb7[ < ` T x   :  6  q   ] / S  7 [   5 ~  X|2V 0y Sw-Q+tNr(L&oIm#G!jDhB b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dI[$m36o*He Z#l  5 ~ G   Y  "k4 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  TT)<*<44>$?$IIST^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$ | |""$h%h'')P*P,,.8/8113 4 6689;|;|==?d@dBBDLELGGI4J4LLNOQQSSUnVnXXZR[R]]_6`6bbdeggiikplpnnpXqXssu>""  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLSSTS{!S{!S$S$S/:S(/:S~S~S[S[SS(SqSqSSS S SS SkS kSSL S1CS 1CSQS QSZS ZSCSCSoSoSSSSDSSS/lS /lS5S5SS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 I bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 I bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 I  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`(Im<`d#G26Zw>,Imq:? b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$` D@d$` D@d bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1<x4p } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  SS)?*?44>)?)IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccWW K K  ??33##  ""$q%q'')]*],,.I/I113545668!9!;;= > @@BBDiEiGGIQJQLLN9O9QQS!T!VVX Y [}[}]]_e`ebbdMeMggi5j5llnoqqstvyvyxxza{a}}II11qq b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYS;˃S;˃SSLSSTS!S!S$S$S:S(:SSSaSaSS(SwSwS+S+SSSS SqS qSSL S7CS 7CSmRS mRS[S [SSSSSSSSDSSS#lS #lS S S߉S0߉SYS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b. b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 K_ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b. b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 K_ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b.  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 K_  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`!E}j=aw8rY} 0Tmu'Lp b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@'d&J Im /l.RQu bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1@}:w } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  NN)8*844> ? IIST^y^yhhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''ss] ]""$G%G'')1*1,,./11346z6z88:d;d==?L@LBBD2E2GGIJLLNNPqQqSSUWVWXXZ=[=]]_#`#bbd e g|g|iikblbnnpHqHssu.v.xxz{}}mmSSʼnŋ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSʃSʃSSLSSTS{!S{!S$S$S):S():S~S~S[S[SS(SqSqSSSSSS SkS kSSL S1CS 1CSRS RS[S [S3S3SoSoSmSmSSDSSSlS lS S S߉S0߉S;S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 I bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 I bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 I  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$Hnr#,Pt!E 5Y}Cg7ae?c b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$` D@d $ ` D  @ d  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1,h$` } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  OO)7*744>?IIST^s^shhrWsW} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33##k k""$S%S'');*;,,.#/#113 4 6688:g;g==?M@MBBD1E1GGIJLLNNPkQkSSUOVOXXZ3[3]]_`bbddfmgmiikQlQnnp5q5ssuvxxzz|o}oSSńņ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lYSA˃SA˃SSLSSTS!S!S%S%S:S(:SSSgSgSS(S}S}S+S+SSSS SwS wSSL S=CS =CSRS RS3\S 3\S{S{SSSySySSDSSS lS  lS{ S{ SK߉S0K߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f2 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 L` bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f2 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 L` bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f2  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 L`  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`9>b4X@bh@di&BGk6Z b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@7t6ZY} ? |  > b  % b  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations14q.k( } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  LL)2*244>?IISS]i^ihhrKsK} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''ooU U""$;%;'')!*!,,./1z1z335`6`88:F;F==?*@*BBD E G}G}IIK_L_NNPAQASSU#V#XXZ[]v]v__aXbXddf:g:iiklnnpprosouuwQxQzz|3}3gg׋׍GG b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃS!SL!SSTS!S!S %S %S5:S(5:S~S~SmSmSS(SSSSSSSS S}S }SSL SCCS CCSRS RS[S [S!S!SSSSS/SD/SSSAlS AlSSS_S0_SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 K  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 K  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 K   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`+ImBfm<yCg5=Y}Ae/Sr b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$kAe;X|.R(oEi bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1GU } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  UU)G*G44>9?9IIS%T%^^h i s~s~} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccUU G G  99++  "#%z%z'')l*l,,.^/^113P4P668B9B;;=4>4@@B&C&EEGHJJLMOOQQSmTmVVX[Y[[[]I^I``b5c5eeghjjl m o|o|qqsbtbvvxHyH{{}.~.llތގPP‘“ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S5˃S5˃SSLSSTS!S!S%S%S:S(:S S SSSS(SSSSSSSS SS S SL  SUCS UCSRS RS[S [S'S'SSSSS#SD#SSS5lS 5lS/S/S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W# b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 VP. bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W# b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 VP. bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W#  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 VP.  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`&Jp@(Nr(L,Pv .4XTx @d2V| b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$i;_ 1vHl>Uy'K bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1E O } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  SS)=*=44>(?(IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&&r r""$\%\'')F*F,,.0/011346689;y;y==?c@cBBDMEMGGI8J8LLN O QQSTV|V|XXZd[d]]_L`Lbbd4e4ggijllnoqxqxssu`v`xxzH{H}}00xxbb b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃS!SL!SSTS!S!S %S %S/:S(/:S~S~SmSmSS(SSSSS S SS S}S }SSL SCCS CCSRS RS[S [SESESŴSŴSSSSDSSS/lS /lSSS߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 J bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 J bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 J  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`'KMqt)M 0W{?<`9Cl(Os b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@?\2 V  , s  I m  C   bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1c*q8 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  QQ)?*?44>-?-IIST^^hhrisi} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccQQ ? ?  --  n n""$\%\'')J*J,,.8/8113&4&6689;;=>@y@yBBDgEgGGIUJULLN?O?QQS)T)VVXY[[]]_r`rbbdXeXggi>j>lln"o"qqstvuvuxxzW{W}}99ooQQ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S5˃S5˃SSLSSTS!S!S%S%S:S(:SSSSSS(SSS%S%SSSS SS S SL  SUCS UCSRS RSO[S O[SSSSSSSSDSSS#lS #lS S S߉S0߉S_S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[' b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 WO0 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[' b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 WO0 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1['  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 WO0  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`:}@d>w]s7[a5k;_  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@b4X * o A e  7 | N r  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1Yc(m } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  OO)9*944>#?#IIST^z^zhhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((tt^ ^""$H%H'')2*2,,./11346{6{88:e;e==?O@OBBD9E9GGI#J#LLN O Q|Q|SSUbVbXXZH[H]]_.`.bbdeggiikhlhnnpLqLssu0v0xxz{}}jj܄܆PPÉË66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS'SL'SSTS!S!S/%S/%S5:S(5:S~S~SSS)S()SSSSSSS S  SS SSL SgCS gCSRS RS \S  \SQSQSSSSS;SD;SSSMlS MlSSSqS0qSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  a@ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  a@ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  a@  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Im .Jnx  $HRv#G$.Rn ._Im! b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$vbNr:^&J6"t bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1R$vH } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  SS)C*C44>3?3IIST^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33## { {""$k%k'')[*[,,.K/K113;4;668+9+;;=>@@B C EEGGIsJsLLN_O_QQSKTKVVX7Y7[[]#^#``bceeggiqjqlln]o]qqsItIvvx5y5{{}!~! oo[[ёѓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS/˃S/˃SSLSSTS!S!S%S%S:S(:SSSsSsS S( SSSSS}S}SS SS SSL SICS ICSSS SS?\S ?\SSSSSSS/SD/SSSAlS AlSYSYS)S0)SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vL b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 VT" bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vL b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 VT" bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vL  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 VT"  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`+Ov46Z_Dhj(-Qx68\a b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$t \Dh,P8 pX| bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1P p@ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  TT)B*B44>2?2IIS T ^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffTT B B  00  q q""$_%_'')M*M,,.;/;113)4)6689;;=>@|@|BBDjEjGGIZJZLLNHOHQQS6T6VVX$Y$[[]^``bcewewggiejellnNoNqqs2t2vvxy{{}}llނބPP‡‰44 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS!SL!SSTS!S!S/%S/%S5:S(5:S~S~SSS)S()SSSSSSS S  SS SSL SgCS gCSRS RS\S \S]S]S!S!S9S9S#SD#SSS5lS 5lS)S)S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  `> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  `> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  `>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Im*Nv4Im^ImDhIm,Px6` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@`Lp8 \ $ H  4 r ^ Jn bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1l>b } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  OO);*;44>'?'IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''uua a""$M%M'')9*9,,.%/%11346688:s;s==?_@_BBDKEKGGI7J7LLNOQQSTV{V{XXZc[c]]_K`Kbbd3e3ggijllnoqwqwssu_v_xxzG{G}}//ss[[ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS/˃S/˃SSLSSTS!S!S%S%S:S(:S S SySySS(SSSSSSSS SS SSL SOCS OCS1RS 1RSm[S m[SSS嵅S嵅S'S'SSDSSS/lS /lS S S߉S0߉SwS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xP b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 WU$ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xP b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 WU$ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xP  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 WU$  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`;Bf2<X| ..RY}*(/S& b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@d(L 4  l  T x > b & J 2 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1`0P } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  NN)8*844>$?$IIST^^hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddNN 8 8  ""  kkU U""$?%?''))*),,./11335r6r88:\;\==?F@FBBD0E0GGIJLLNOQ{Q{SSUeVeXXZO[O]]_9`9bbd#e#ggi j llnnpfqfssuJvJxxz.{.}}hhډڋLL00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS-SL-SSTS!S!S%S%S;:S(;:S~S~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CSRS RS[S [SSSgSgSOSOS/SD/SSSAlS AlSkSkS;S0;SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 I bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 I bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 I  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Im 1Nr2Vqc):Gk-Q+OlUy b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$&(*, . 02 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1]:t } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  QQ)?*?44>-?-IIST^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccQQ ? ?  --  n n""$\%\'')J*J,,.8/8113&4&6689;;=>@y@yBBDgEgGGIUJULLN?O?QQS)T)VVXY[[]]_r`rbbd\e\ggiFjFlln0o0qqstvvxy{y{y}}cc؂؄MM‡‰77!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS#˃S#˃SSLSSTS!S!S%S%S:S(:S~S~SgSgSS(S}S}SSSwSwSS SwS wSSL S=CS =CS SS  SSE\S E\SSSISISISIS5SD5S S SGlS GlSkSkS;S0;SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kA  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 IK bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kA  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 IK bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kA   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 IK  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`0T| D@p46Z\ JnPt<`^ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$"} {ywusq bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1[6l } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  ^^)J*J44>6?6IIST^^hiszsz} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jj^^ R R  FF::..""  "#%%'(*x*x,,.d/d113P4P668<9<;;=(>(@@BCEEGHJvJvLLN^O^QQSFTFVVX.Y.[[]^``bbdrerggiZjZllnBoBqqs*t*vvxy{{}}nnVVʇʉ>>&& b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS'SL'SSTS!S!S%S%S;:S(;:S~S~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CSRS RS'\S '\SoSoSwSwSSSSDSSS/lS /lS S S߉S0߉SwS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 I bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 I bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 I  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Im-Qx8Eie%AeLp =a9] Dq b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@4 6  8  :  <  > @B bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1Q. hE } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP):*:44>$?$IIS T ^}^}hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((vvb b""$M%M'')7*7,,.!/!113 4 6688:j;j==?T@TBBD>E>GGI(J(LLNOQQSSUgVgXXZM[M]]_3`3bbdeggiikrlrnnpXqXssu>v>xxz${$}} }}cc։֋II// b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS)˃S)˃SSLSSTS!S!S%S%S:S(:SSSgSgSS(S}S}SSS}S}SS SwS wSSL S=CS =CSCRS CRS[S [SǚSǚS;S;SSSSDSSS/lS /lS S S߉S0߉SkS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1mE b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 IL bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1mE b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 IL bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1mE  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 IL  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`<Dh5:Sw5 %IQu5#G5 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@o m  k  i g e ca bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1G"}X3 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  XX)@*@44>(?(IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhXX H H  88((  ""$l%l'')T*T,,.>"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS-SL-SSTS!S!S %S %SA:S(A:S~S~SmSmSS(SSSSSSSS S}S }SSL SCCS CCS RS  RSI[S I[SSSSSgSgS/SD/SSSAlS AlSMSMSS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 J bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 J bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 J  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`4Tx~$Dhrx4Xfh$HZ~X| b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$ .8B(L2V<`Fj bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1aB# } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  WW)C*C44>-?-IIST^^hhrqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1gg[[ O O  CC77''  ""$u%u'')a*a,,.M/M113949668%9%;;=>@@BBDmEmGGIUJULLN=O=QQS%T%VVX Y [[]]_i`ibbdQeQggi9j9lln!o!qqs t v}v}xxze{e}}MM55uu b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS˃S˃SSLSSTS!S!S %S %S:S(:S~S~SmSmSS(SSS S SqSqSS S}S }SSL SCCS CCS?SS ?SS{\S {\SÛSÛSSSySyS#SD#SSS5lS 5lS S S߉S0߉SkS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`6 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 IO bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`6 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 IO bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`6  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 IO  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Im5Y%I(L$'Kmt>Gks=f 0 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$8(L<`Ptdx&: bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1fL2 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  RR)<*<44>$?$IIS T ^}^}hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++wwa a""$K%K'')5*5,,./113 4 6~6~88:h;h==?P@PBBD6E6GGIJLLNOQuQuSSU[V[XXZA[A]]_'`'bbd e ggiikflfnnpLqLssu2v2xxz{}}qqWWɉɋ;; b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS-SL-SSTS!S!S %S %SA:S(A:S~S~SmSmSS(SSSSSSSS S}S }SSL SCCS CCSRS RS9\S 9\SSSSSSSSDSSS/lS /lS S S߉S0߉SMS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 J bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 J bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 J  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`0Tpz<Jnl .$HTx"Fjl.R b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@PtZ ~  d n  x ! +5? bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1eF'i } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  SS);*;44>!?!IIST^w^whhr[s[} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggWW G G  77''o o""$W%W'')?*?,,.'/'11346688:k;k==?Q@QBBD5E5GGIJLLNNPoQoSSUSVSXXZ7[7]]_`bbddfqgqiikUlUnnp9q9ssuvxxz{}s}sWWɄɆ;; b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS#˃S#˃SSLSSTS!S!S %S %S:S(:S~S~SmSmSS(SSSSSwSwSS S}S }SSL SCCS CCS}QS }QSZS ZSSSSSSSSDSSS#lS #lS S Sc߉S0c߉S S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b: b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 IN bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b: b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 IN bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b:  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 IN  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Im=Ei#G;Tx! &JjRvFj$H b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@* N > b R v f  z (<,P bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1~dJ0 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  PP)6*644>?IISS]m^mhhrOsO} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++ssY Y""$?%?'')%*%,,. / 1~1~335d6d88:J;J==?.@.BBDEGGIIKcLcNNPEQESSU'V'XXZ [ ]z]z__a\b\ddf>g>iik l nnpqssssuuwUxUzz|7}7kkۋۍKK b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃS3SL3SSTS!S!S%S%SG:S(G:S~S~SsSsS S( SSSSS#S#SS SS SSL SICS ICSRS RSC[S C[SSSQSQSSSGSDGSSSYlS YlSSSqS0qSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 J bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 J bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 J  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`7yZ~ 3SwVz/S 0Sw /,Osx b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$ D@d`40TPtp bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1lXD0 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  YY)K*K44>=?=IIS)T)^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggYY K K  ==//!!  "#%~%~'')p*p,,.b/b113T4T668F9F;;=8>8@@B*C*EEGHJJL M OOQQSqTqVVX_Y_[[]M^M``b9c9eeg#h#jjl m ooqqsftfvvxLyL{{}2~2ppTTƑƓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqS˃S˃S{SL{SSTS!S!S%S%S:S(:S~S~SsSsS S( SSSSSkSkSS SS SSL SICS ICSRS RS-\S -\SuSuSSSSS5SD5S S SGlS GlSASASS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U+ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 IV bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U+ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 IV bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U+  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 IV  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Im2Vx#G8j!*jLp Dh>b Dv b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$*NTx ~7=ag &J bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1qbSD } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  WW)A*A44>,?,IIST^^hhrrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjZZ J J  ::** v v""$`%`'')J*J,,.4/411346689;};}==?g@gBBDQEQGGI1?1IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggUU C C  11  r r""$`%`'')N*N,,.@}@}BBDkEkGGIYJYLLNCOCQQS-T-VVXY[[]^`v`vbbd\e\ggiBjBlln&o&qqstvyvyxxz[{[}}==ssUU b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqS˃S˃SSLSSTS!S!S%S%S:S(:S~S~SsSsS S( SSSSSkSkSS SS SSL SICS ICSQS QSZS ZSSSSS_S_S)SD)SSS;lS ;lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W/ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 IW bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W/ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 IW bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W/  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 IW  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Im67[w)M%Aef - /0TzpZ~ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@P t z  3 9 ] c "FLpv / bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations15&y } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  SS)=*=44>'?'IIS T ^~^~hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,xxb b""$L%L'')6*6,,. / 113 4 6688:i;i==?S@SBBD=E=GGI'J'LLN O QQSSUfVfXXZL[L]]_2`2bbdeggiiklllnnpPqPssu4v4xxz{}}nnTTljNj::  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃSQSQSS$S_:S_:S~S~SaS$aSשSשS;S;SS$SQSQSZS4ZSST Se S e S_S _S7S 7SqlS qlSS SSS=S} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Qa b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Qa b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Qa b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 mGk!Eh  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10xx  hh)X*X44>F?FIIS2T2^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1xxhh X X  HH88((  "#%%'')p*p,,.`/`113P4P668@9@;;=0>0@@B C EEGHJJLLNrOrQQS^T^VVXJYJ[[]6^6``b"c"eeghjjllnpopqqs\t\vvxHyH{{}4~4  nn b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS˃S˃SuS uSSS:S:S~S~SISISSpSeSeSSSQS`QSZSZSS$ Se S e SMS MS%S %S_lS _lSwS wSGSGSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1e} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1e} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I} b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1e k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Im#G!jDh  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  aa)O*O44>=?=IIS+T+^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ssaa O O  ==++ ~ ~""$l%l'')Z*Z,,.H/H113646668$9$;;=>@@BCEwEwGGIeJeLLNSOSQQSATAVVX/Y/[[]^``b c eeggipjpllnYoYqqs=t=vvx!y!{{}~ww[[͇͉??## b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃSKSKSS$S_:S_:S~S~SaS$aSשSשS;S;SS$SQSQSZS4ZSST Se S e SGS GSS SYlS YlSMS MSSSS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R_ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R_ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R_ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 mGk!Eh  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10vv  bb)N*N44>8?8IIS T ^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvbb N N  ::&&t t""$`%`'')L*L,,.8/8113$4$6689;;==?r@rBBD^E^GGIHJHLLN0O0QQSTVVXY[t[t]]_\`\bbdDeDggi,j,llnoqqssupvpxxzX{X}}@@((ll b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS˃S˃S{S {SSS:S:S~S~SCSCSSpSeSeSSSQS`QSZSZSS$ Se S e S;S ;SS SMlS MlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I}! b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1d} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I}! b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1d} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1I}! b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1d k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Im#G!jDh  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  ]])G*G44>1?1IIST^^hhrqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ss]] G G  11zzd d""$N%N'')8*8,,."/"113 4 6688:k;k==?U@UBBD?E?GGI)J)LLNOQQSSUrVrXXZ\[\]]_F`Fbbd0e0ggijllnoqsqsssuWvWxxz;{;}}uuYYˎː== b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃSES ESSSY:SY:S~S~S[S[SѩSpѩS5S5SSSQS`QSZSZSS$ Se S e SMS MS%S %S_lS _lSS SSSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 oIm#G!j  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  bb)P*P44>>,,  ""$m%m'')[*[,,.I/I113747668%9%;;=>@@BCExExGGIdJdLLNNONQQS8T8VVX"Y"[[] ^ ``bbdkekggiUjUlln?o?qqs)t)vvxy{{}}rr\\чщFF00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS˃S˃S{S {SSS:S:S~S~S1S1SSpSkSkSSSQS`QSZSZSS$ Se S e SGS GSS SYlS YlS}S }SMSMSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R( b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1V} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R( b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1V} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R( b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1V k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Im#G!jDh  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  gg)S*S44>=?=IIS%T%^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ssgg [ [  OOCC77++  "#%%' ( **,,.m/m113Y4Y668E9E;;=1>1@@BCEEG H J}J}LLNeOeQQSMTMVVX5Y5[[]^``bceyeyggiajallnIoIqqs1t1vvxy{{}~uu]]чщEE-- b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃSES ESSSS:SS:S~S~S[S[S˩Sp˩S/S/SSSQS`QSZSZSS$ Se S e S5S 5S S  SGlS GlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 oIm#G!j  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  __)I*I44>1?1IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ss__ K K  77##q q""$\%\'')F*F,,.0/011346689;y;y==?c@cBBDMEMGGI5J5LLNOQQSTVtVtXXZZ[Z]]_@`@bbd&e&ggi j llnnpeqessuKvKxxz1{1}}ppVVɎɐ<< b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS˃S˃SS SSS:S:S~S~S7S7SSpSkSkSSSQS`QSZSZSS$ Se S e S5S 5S S  SGlS GlS S  S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R* b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1W} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R* b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1W} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R* b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1W k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Im#G!kEi  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  cc)K*K44>1?1IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1sscc S S  CC33##  "#%w%w'')_*_,,.G/G113/4/6689;;==?s@sBBD[E[GGIAJALLN%O%QQS T V{V{XXZ_[_]]_C`Cbbd'e'ggi j l}l}nnpaqassuEvExxz){)}} ccՉՋGG++ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃS?S ?SSSS:SS:S~S~SaSaS˩Sp˩S/S/SSSQS`QS+[S+[SS$ S_ S _ S;S ;SS SMlS MlSYS YS)S)SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ImBf<`6Z  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  dd)P*P44>8?8IIS T ^^hiszsz} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1tthh \ \  PPDD44   " # %%'')n*n,,.Z/Z113F4F668292;;=>@@BCExExGGI`J`LLNHOHQQS0T0VVXY[[]^`t`tbbd\e\ggiDjDlln,o,qqstvvxxzp{p}}XX̂̄@@%% ~~ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS˃S˃SS SSS:S:S~S~S7S7S Sp SqSqSSS%RS`%RSa[Sa[SS$ S_ S _ S)S )SS S;lS ;lS S  S߉S߉SqS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[1 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1_} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[1 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1_} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[1 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1_ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 !ztn  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  WW)A*A44>'?'IIS T ^^hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1llXX D D  00||f f""$P%P''):*:,,.$/$11346688:m;m==?S@SBBD9E9GGIJLLNOQxQxSSU^V^XXZD[D]]_*`*bbdeggiikilinnpOqOssu5v5xxz{}}ttXXʉʋ<<  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃS?S ?SSSM:SM:S~S~SaSaSũSpũS)S)SSS1RS`1RSm[Sm[SS$ S_ S _ S#S #SS S5lS 5lS S  S߉S߉SYS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ImUya m  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10rr  ^^)F*F44>*?*IIST^^hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1rrbb R R  BB22 z z""$b%b'')J*J,,.2/211346689;v;v==?Z@ZBBD>E>GGI"J"LLNOQxQxSSU\V\XXZ@[@]]_$`$bbdegzgziik^l^nnpBqBssu&v&xxz { }|}|``҄҆BB$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS˃S˃SS SSS:S:S~S~S7S7S Sp SqSqSSS1RS`1RSm[Sm[SS$ S_ S _ SS SS S)lS )lS S  Si߉Si߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[3 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1^} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[3 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1^} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[3 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1^ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ,( y  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  WW)=*=44>?IIST^r^rhhrRsR} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nnZZ F F  22zz` `""$F%F''),*,,,./11335k6k88:Q;Q==?3@3BBDEGGIIKhLhNNPJQJSSU,V,XXZ[]]__aabaddfCgCiik%l%nnpqsxsxuuwZxZzz|<}<nnދލNN b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS9S 9SSSM:SM:S~S~SgSgSũSpũS)S)SSSQS`QS[S[SS$ SY S Y SGS GSS SYlS YlSS SqSqSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Im8\2V,P  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  bb)T*T44>D?DIIS0T0^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ppbb T T  FF88**  "#%%'(*y*y,,.k/k113]4]668O9O;;=A>A@@B3C3EEG%H%JJLMOOQRTxTxVVXfYf[[]T^T``b@c@eeg*h*jjlmooqqsmtmvvxSyS{{}9~9uuYYˑ˓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS#˃S#˃SS SSS:S:S~S~S%S%SSpSwSwSSS+RS`+RSg[Sg[SS$ SY S Y S/S /SS SAlS AlS;S ;S S SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d: b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Q} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d: b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Q} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d: b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Q k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ("{u  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  XX)B*B44>+?+IIST^^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kk[[ K K  ;;++ w w""$a%a'')K*K,,.5/51134668 9 ;~;~==?h@hBBDRERGGI;J;LLN#O#QQS T VVXXZg[g]]_O`Obbd7e7ggijllnoq{q{ssucvcxxzK{K}}33yycc b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS3S 3SSSG:SG:S~S~SgSgSSpS#S#SSSRS`RS!\S!\SS$ SY S Y S/S /SS SAlS AlS/S /S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Im )Mj .  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  \\)J*J44>6?6IIST^^hispsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nn\\ J J  88&& y y""$g%g'')U*U,,.C/C1131416689;;= > @@BBDrErGGI^J^LLNHOHQQS2T2VVXY[[]^`{`{bbdaeaggiGjGlln+o+qqs t v~v~xxz`{`}}BB$$vvXX b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS#˃S#˃SS SSS:S:S~S~S%S%SSpSwSwSSSRS`RS[S[SS$ SY S Y S#S #SS S5lS 5lS S  S߉S߉SkS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d< b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1R} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d< b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1R} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d< b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1R k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  D4XHl\  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  VV)@*@44>(?(IIS T ^^hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kkWW C C  //{{e e""$O%O'')9*9,,.#/#113 4 6688:l;l==?V@VBBD@E@GGI(J(LLNOQQSSUgVgXXZM[M]]_3`3bbdeggiikmlmnnpQqQssu5v5xxz{}}ooSSƉƋ99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS3S 3SSSG:SG:S~S~SaSaSSpS#S#SSSQS`QS [S [SS$ SY S Y SAS ASS SSlS SlSS S}S}S%S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Im.R(L"F  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  \\)L*L44>:?:IIS&T&^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ll\\ L L  <<,,   ""$t%t'')d*d,,.T/T113D4D668494;;=$>$@@BCEEGHJzJzLLNfOfQQSRTRVVX>Y>[[]*^*``bceeghjxjxllndodqqsPtPvvx1?1IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggUU C C  11  r r""$`%`'')N*N,,.@}@}BBDkEkGGIYJYLLNGOGQQS5T5VVX#Y#[[]^``bbdvevggidjdllnMoMqqs1t1vvxy{{}}kk݂݄OO33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS-S -SSSA:SA:S~S~SaSaSSpSSSSS SS` SSE\SE\SS$ SY S Y S)S )SS S;lS ;lS5S 5SSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Im=a/S  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  VV)B*B44>,?,IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjVV B B  ..||h h""$T%T'')@*@,,.,/,11346689;z;z==?f@fBBDRERGGI%?%IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggQQ ; ;  %%nnX X""$B%B''),*,,,./11346u6u88:_;_==?I@IBBD3E3GGIJLLNOQ|Q|SSUfVfXXZP[P]]_:`:bbd$e$ggijllnnpgqgssuKvKxxz/{/}}iiۉۋMM11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS-S -SSS;:S;:S~S~S[S[SSpSSSSSQS`QS[S[SS$ SS S S S/S /SS SAlS AlSeS eS5S5SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Jn$H?9  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  VV)D*D44>0?0IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhVV D D  22  s s""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIXJXLLNBOBQQS,T,VVXY[[]^`u`ubbd_e_ggiIjIlln3o3qqstvvxy{|{|}}ffۂۄPPŇʼn::$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS/˃S/˃SS SSS:S:SSSSSSpS}S}SSSRS`RS[[S[[SS$ SS S S S/S /SS SAlS AlSeS eS5S5SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vL b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1N} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vL b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1N} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vL b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1N k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 & ys  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  [[)G*G44>1?1IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1gg[[ O O  CC77++  "#%%'')u*u,,.a/a113M4M668999;;=%>%@@BCEEGGIqJqLLNYOYQQSATAVVX)Y)[[]^``bbdmemggiUjUlln=o=qqs%t%vvx y {{}}ii݂݄QQŇʼn99!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS'S 'SSS;:S;:S~S~S[S[SSpSSSSSSS`SSK\SK\SS$ SS S S SS SS S/lS /lS S  S߉S߉SwS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ImAe 9]  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  SS)=*=44>%?%IIS T ^~^~hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++yye e""$P%P''):*:,,.$/$11346688:m;m==?W@WBBDAEAGGI)J)LLNOQQSSUhVhXXZN[N]]_4`4bbdeggijlslsnnpYqYssu?v?xxz%{%}} ~~dd׉׋JJ00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS/˃S/˃SS SSS:S:S S SSSSpSSSSSSS`SS\S\SS$ SS S S SS SS S)lS )lS S  S߉S߉SeS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vN b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1O} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vN b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1O} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vN b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1O k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 \2r$H  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  WW)?*?44>%?%IIS T ^{^{hhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggWW G G  77''  ""$k%k'')S*S,,.;/;113#4#668 9 ;;==?g@gBBDOEOGGI5J5LLNOQQSSUoVoXXZS[S]]_7`7bbdeggiikqlqnnpUqUssu9v9xxz{}}ssWWɉɋ;; b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃS'S 'SSS5:S5:S~S~SUSUSSpSSSSSRS`RS[S[SS$ SM S M S#S #SS S5lS 5lSAS ASSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 :,PBfX|  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  XX)D*D44>,?,IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hh\\ P P  DD88((  "#%v%v'')b*b,,.N/N113:4:668&9&;;=>@@BBDlElGGITJTLLN?IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``LL 8 8  $$ppZ Z""$D%D'').*.,,./11346w6w88:a;a==?G@GBBD-E-GGIJLLNNPlQlSSURVRXXZ8[8]]_`bbdeggiikglgnnpMqMssu3v3xxz{}}rrVVȉȋ:: b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃS!S !SSS5:S5:S~S~SUSUSSpSSSSSQS`QSZSZSS$ SM S M S S  SS SlS lS S  S߉S߉SAS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 nJn&J&p  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  RR):*:44>?IIST^x^xhhrZsZ} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&&n n""$V%V'')>*>,,.&/&11346688:j;j==?N@NBBD2E2GGIJLLNNPlQlSSUPVPXXZ4[4]]_`bbdegrgriikVlVnnp:q:ssuvxxz{}t}tXXʄʆ:: b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S5˃S5˃SS SSS:S:S S SSSSpSSSSSQS`QSZSZSS$ SM S M SS SS SlS lS S  SW߉SW߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W# b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1X} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W# b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1X} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W# b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1X k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 SwAe/SA  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  KK)1*144>?IISS]n^nhhrNsN} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&nnT T""$:%:'') * ,,./1y1y335_6_88:E;E==?'@'BBD E GzGzIIK\L\NNP>Q>SSU V XXZ[]s]s__a]b]ddf?g?iik!l!nnpqststuuwVxVzz|8}8jjڋڍJJ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LYSʃSʃSS SSS/:S/:S~S~SOSOSSpS S SSSQS`QS[S[SS SM S| M S;S ;Sc~S c~SANS ANSS SމSމS3S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1,4<  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  VV)J*J44>:?:IIS&T&^^hiss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddVV H H  ::,,  "#%}%}'')o*o,,.a/a113S4S668E9E;;=7>7@@B)C)EEGHJJL M OOQQSnTnVVX\Y\[[]J^J``b6c6eeg h jjl m ooqqsotovvxUyU{{};~;!!ww[[͓͑ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYS5˃S5˃SS SSS:S:SSSSS%Sp%SSSSS%RS`%RSa[Sa[SS SM S| M S)S )SQ~S Q~S/NS /NS5S 5S+މS+މSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^* b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1K} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^* b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1K} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^* b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1K k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1o&Fjl#  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  LL)@*@44>)?)IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__OO ? ?  //  k k""$_%_'')I*I,,.3/311346689;|;|==?f@fBBDPEPGGI9J9LLN!O!QQS T V}V}XXZe[e]]_M`MbbdAeAggi)j)llnoqqssumvmxxzU{U}}==%%mm b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYSʃSʃSS SSS):S):S~S~SUSUSSpSSSSSQS`QSZSZSS SM S| M S#S #SK~S K~S)NS )NS#S #SމSމSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1J~ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1o9%I  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  PP)B*B44>.?.IIST^^histst} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$_%_'')M*M,,.;/;113)4)6689;;=>@|@|BBDjEjGGIVJVLLN@O@QQS*T*VVXY[[]]_s`sbbdYeYggi?j?lln/o/qqstvvxxzd{d}}FF(( zz\\ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LYS;˃S;˃SS SSS:S:SSSSS%Sp%SSSSS}QS`}QSZSZSS SM S| M SS SE~S E~S#NS #NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`, b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1L} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`, b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1L} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`, b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1L k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1Im4X/SUy  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  JJ)<*<44>$?$IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__KK 7 7  ##ooY Y""$K%K'')5*5,,./113 4 6~6~88:h;h==?R@RBBDsHl,Qu6Z/S$Hm 16|[ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Ko)M+v Tx2V4];_= f  D h  " F   $ o o   M q 5 Y  +O ^ -$Hx VzNr4X8w6=a_=a?hFj$H&qOs-Q b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dKa,wB X#n9B   O @    e 0  {  F\ bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10^^ LL)D*D44>4?4IIS"T"^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1^^LL : : qq__{{i i""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGI\J\LLNJOJQQS8T8VVX&Y&[[]^``bceyeyggigjglln\o\qqs@t@vvx$y${{}~zz^^ЇЉBB&&# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSM˃SM˃SSdSSlS{!S{!S$S$S:S@:S)S)S[S[SS@SqS qS=S=SSSS SkS kSSd S1CS 1CSQS QS[S [SOSOSeÅSeÅS!S!S+S+S)&S )&S5S5S Sp SGlStGlSySTySSSqS$qSS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|H b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 K  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|H b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 K  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|H  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 K   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`!EKoVz9?cJn -3W>b!'K2Vw b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Ko)M+v Tx2V4];_= f  D h  " F   $ o p   M q 6 Z  +O _ -%Ix VzNr4X8w6=a_=a?hFj$H&qOs-Q b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dKa,wB X#n9E   O D    e 0  {  F\ bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  SS)*44>5?5IIS!T!^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33## { {""$m%m'')]*],,.M/M11346688:w;w=>@@B C EEGGIuJuLLNaOaQQSMTMVVX9Y9[[]%^%``bceeggisjsllnkokqqsWtWvvxCyC{{}/~/}}iiߑߓ  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1SS{ʃS{ʃSS|SSSSS"S|"S&SH&S9S9SW~SPW~SSXS{S{SSPSSSkSkSS$ SuS, uSS SS$ S S  SQDS| QDSRSXRS[S|[S S SeɅSxeɅS!S\!S S S (S (SS8SSS/lS/lSySyS Sl S߉S߉SwS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O| b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1; bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O| b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1;  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O|  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1; bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$Hk>Aee2V;;_Jn5b b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@dt  0@,P<`Lp\ l   |  (   8  $ H  4XDhTxb n'zf$,P 0Uy<?~$HDh0T /n<`HlTx` lx" .  !:!!! b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1ddH,tX< hJ ,  >  p }  R <  4{ :xZ< bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10^^ HH):*:44>&?&IIST^^hisrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1^^HH 2 2 x xbbLLeeO O""$A%A'')+*+,,./11335t6t88:^;^==?H@HBBD2E2GGIJLLNOQ}Q}SSUgVgXXZQ[Q]]_;`;bbd%e%ggijllnoqtqtssuXvXxxz<{<}} vvZZ̎̐>># b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSS˃SS˃SSdSSlS{!S{!S$S$S:S@:S)S)S[S[SS@SqS qS=S=SSSS SkS kSSd S1CS 1CSQS QS%[S %[SmSmSeÅSeÅS!S!S+S+S)&S )&S#S#SSpS5lSt5lSySTyS)S)S߉S$߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 K  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 K  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 K   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Y}$)MdSw.RdNr6ZZ~Hl, b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Ko)M+v Tx2V4];_= f  D h  " F   $ o   M q  +O -x Vz4X6'_f=a,P?Uy?h~FjDh$H .n&qOs-Q b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dKa,wB X#n9  O   e B 0  @  {   F>\ bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  MM)*44>)?)IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%ss_ _""$O%O'');*;,,.'/'1n1n335Z6Z88:F;F==?a@aBBDMEMGGI9J9LLN!O!QQS T V}V}XXZe[e]]_M`Mbbd5e5ggijllnoqqssumvmxxzU{U}}==%% ii  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSʃSʃSSdSSlS{!S{!S$S$S/:S@/:S~S~S[S[SS@SqS qSSS S SS SkS kSSd S1CS 1CSRS RS'\S '\SoSoSkÅSkÅS'S'S1S1S/&S /&S)S)SSpS;lSt;lSySTyS_S_S/S$/SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 K bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 K bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 K  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`=Fj,P'0Tx>`b(Lpy.R b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Ko)M+v Tx2V4_=a? h  F j  $ H   & . R q    O s Q u  -Q7t /6Zz X|Y}6Z?|8a?cAjHl&J(sQu/S b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dKa,wB X#no9, i  O &   c e 0  {  F\ bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  VV)B*B44>.?.ItItST^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbVV J J  >>22&&  "#%%'')p*p,,.\/\113H4H668494;;= > @@B C EEGGInJnLLNVOVQQS>T>VVXXZl[l]]_T`TbbdjejggiRjRllnDoDqqs,t,vvxy{{}}ppXẊ̉@@((  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSG˃SG˃SSdSSlS{!S{!S$S$S:S@:S#S#S[S[SS@SqS qS7S7SSSS SkS kSSd S1CS 1CSIRS IRS[S [S͚S͚SkÅSkÅS'S'S1S1S/&S /&S#S#SSpS5lSt5lSySTySYSYS)S$)SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q= b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 K] bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q= b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 K] bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q=  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 K]  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ko DTx4cDh$4X|x .$Hh"F b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Ko)M+v Tx2V4];_= f  D h  " F   $ # G o   M q F j  ,P,i .+Oy W{Nr5Y4q7`>b@iGk%I'rPt.R b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dKa,wB X#nf9# `  O    Z e 0  {  F\ bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  QQ)?*?44>-?-IIST^^hhrxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccQQ ? ?  --  n n""$\%\'')J*J,,.8/8113&4&6689;;=>@y@yBBDgEgGGIUJULLN?O?QQS)T)VVXY[[]]_r`rbbd\e\ggiFjFlln2o2qqstvvxy{M{M}}77!!ćĉ99##  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSʃSʃSSdSSlS{!S{!S$S$S):S@):S~S~S[S[SS@SqS qSSSSSS SkS kSSd S1CS 1CSQS QSZS ZSSSkÅSkÅS'S'S1S1S/&S /&SSSSpS#lSt#lSySTyS S S߉S$߉S_S(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 K bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 K bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 K  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$o'Kh!Mq8\h+O,>bp -8x b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Ko)M+v Tx2V4];_= f  D h  " F   $ o   M q  +O -x Vz4X6>b$_a?c#GA Fj,jiHl+O&J(sQu/S b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dKa,wBZ%p;  Q   g ] 2   } W  H^ bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  RR):*:44>"?"IeIeST^^hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbRR B B  22"" z z""$f%f'')N*N,,.6/611346689;z;z==?b@bBBDJEJGGI2J2LLNOQQSSU=V=XXZ![!]]_`bbdeggiiknlnnnpZqZssu>v>xxz"{"}}xx\\Ή΋@@$$  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSM˃SM˃SSdSSlS{!S{!S$S$S:S@:S#S#S[S[SS@SqS qS7S7SSSS SkS kSSd S1CS 1CSwQS wQSZS ZSSSkÅSkÅS'S'S1S1S/&S /&SSSSpS#lSt#lSySTyS S S߉S$߉SkS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uA b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 K\ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uA b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 K\ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uA  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 K\  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ko&0Tc)M**Ns+&vImm TxC b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Ko)M+v Tx2V4];_= f  D h  " F   $ o   M q  -Q /z X|6Z83WaVz?c>{A=a#j`Hl#G&J (sQu/S b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dKa,wB X#n9  O    g T 2   } N  H ^ bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  NN)8*844>"?"IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&tt` `""$K%K'')5*5,,./113 4 6~6~88:h;h==?R@RBBD?IIST^u^uhhr[s[} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__KK 7 7  ##ooY Y""$C%C'')-*-,,./11346v6v88:`;`==?H@HBBD.E.GGIJLLNNPmQmSSUSVSXXZ9[9]]_`bbdegxgxiik^l^nnpDqDssu*v*xxz{}}ii܄܆OO33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SA˃SA˃SSLSSTS!S!S%S%S:S(:SSSgSgSS(S}S}S+S+SSSS SwS wSSL S=CS =CS+RS +RSg[S g[SSSUSUSSS#SD#SSS5lS 5lS;S;S S0 SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f2 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 KY bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f2 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 KY bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f2  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 KY  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`CRv* />b(*Nm(:Y}& b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@` 2V(m?c5zLpB bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1F P  Z  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  WW)C*C44>-?-IIST^^hhrqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1gg[[ O O  CC77''  ""$u%u'')a*a,,.M/M113949668%9%;;=>@@BBDmEmGGIUJULLN=O=QQS%T%VVX Y [[]]_i`ibbdQeQggi9j9lln!o!qqs t v}v}xxze{e}}MM55uu b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃS!SL!SSTS!S!S$S$S/:S(/:S~S~SaSaSS(SwSwSSS S SS SqS qSSL S7CS 7CSRS RS!\S !\SiSiSSSSSSDSSS lS  lS{ S{ SK߉S0K߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 K bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 K bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 K  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`x$,Pe8*0TsLp(Mq q'K D b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ DFj .l0TVz>| bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 L H  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  JJ)0*044>?IISS]g^ghhrIsI} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%mmS S""$9%9'')*,,./1x1x335^6^88:D;D==?(@(BBD E G{G{IIK]L]NNP?Q?SSU!V!XXZ[]t]t__aVbVddf8g8iiklnnpprmsmuuwOxOzz|1}1eeՋՍEE b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SA˃SA˃SSLSSTS!S!S%S%S:S(:SSSgSgSS(S}S}S1S1SSSS SwS wSSL S=CS =CSRS RS[S [S'S'SSSSS SD SSSlS lS S S߉S0߉S;S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j6 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 KZ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j6 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 KZ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j6  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 KZ  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$3W^B 11Uz<`  /Txx Z~.R x b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Y}+O!f8\ .sEi; bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1  d ) n3 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  QQ)9*944>?IIST^u^uhhrYsY} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%%m m""$U%U'')=*=,,.%/%113 4 6688:i;i==?O@OBBD3E3GGIJLLNNPmQmSSUQVQXXZ5[5]]_`bbddfogoiikSlSnnp7q7ssuvxxzz|q}qUUDŽdž99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃS'SL'SSTS!S!S %S %S;:S(;:S~S~SmSmSS(SSSSSSSS S}S }SSL SCCS CCS%RS %RSa[S a[SSSSSSSSDSSS/lS /lS)S)S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  M  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  M  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  M   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`;KoBfW{$Hj*cLpw  .RoY} b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@` :^8[5Y3| Vz bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1F X ! j  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  OO)9*944>$?$IIS T ^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbRR B B  22""n n""$X%X'')B*B,,.,/,11346689;u;u==?_@_BBDIEIGGI4J4LLNOQQSTVxVxXXZ`[`]]_H`Hbbd0e0ggijllnoqtqtssu\v\xxzD{D}},,tt^^ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S;˃S;˃SSLSSTS!S!S%S%S:S(:SSSSSS(SSS%S%SSSS SS S SL  SUCS UCS3SS 3SSo\S o\SSS_S_SMSMS5SD5S S SGlS GlSSS_S0_SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[' b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 TR. bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[' b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 TR. bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1['  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 TR.  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`BPt+Fj9]{ !Ec&J -Ko}f* b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@4X@(x`Hl0T< bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations14 T $ t  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  YY)K*K44>=?=IIS)T)^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggYY K K  ==//!!  "#%~%~'')p*p,,.b/b113T4T668F9F;;=8>8@@B*C*EEGHJJL M OOQQSqTqVVX_Y_[[]M^M``b9c9eeg#h#jjl m ooqqsftfvvxLyL{{}2~2ppTTƑƓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃS!SL!SSTS!S!S %S %S5:S(5:S~S~SmSmSS(SSSSSSSS S}S }SSL SCCS CCSQS QS [S  [SUSUSc…Sc…SSSSDSSS#lS #lS S S߉S0߉S_S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  L bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  L bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  L  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`W{$2VbIm44X;_6]-Q;_ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@0T .wQu+O)rLp&J bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 3 | E W } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  MM)7*744>!?!IIST^x^xhhr^s^} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&rr\ \""$F%F'')0*0,,./11346y6y88:c;c==?M@MBBD7E7GGI!J!LLNOQzQzSSU`V`XXZF[F]]_,`,bbdeggiikflfnnpJqJssu.v.xxz{}}hhڄچNN44 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S;˃S;˃SSLSSTS!S!S%S%S:S(:SSSSSS(SSS+S+SSSS SS S SL  SUCS UCS+RS +RSg[S g[SSSąSąS[S[SSDSSS/lS /lSSS߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_+ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 UQ0 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_+ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 UQ0 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_+  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 UQ0  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`(<`d'>Bf% Dg#"Ei~! b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$t \Dh,P8 pX| bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1D  d4 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>/?/IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIWJWLLNAOAQQS+T+VVXY[[]]_t`tbbdZeZggi@j@lln$o$qqstvwvwxxzY{Y}};;qqSS b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS-SL-SSTS!S!S/%S/%SA:S(A:S~S~SSS)S()SSSSSSS S  SS SSL SgCS gCSRS RS9\S 9\SSSSS_S_S)SD)SSS;lS ;lSSSSS#S0#SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  c@ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  c@ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  c@  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ko:@dBc7["Bfl DKo%I b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@8\(L<,p `Pt bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations14 \ 0   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  PP)>*>44>.?.IIST^^hisrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$[%['')I*I,,.7/7113%4%6689;;=>@x@xBBDfEfGGIVJVLLNDODQQS2T2VVX Y [[]^``bbdsesggiajallnJoJqqs.t.vvxy{{}}hhڂڄLL00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS/˃S/˃SSLSSTS!S!S%S%S:S(:S S SsSsS S( SSSSSSSS SS SSL SICS ICS SS  SSE\S E\SSSiSiSSS;SD;SSSMlS MlSSSwS0wSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xP b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 TN" bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xP b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 TN" bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xP  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 TN"  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`ANr%:^2Vs :nW{,;_l<` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ DB@><:86 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1" } X 3   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  WW)G*G44>7?7IIS#T#^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggWW G G  77''  ""$o%o'')_*_,,.O/O113?4?668/9/;;=>@@BCEEGGIwJwLLNcOcQQSOTOVVX;Y;[[]'^'``bceeggiujullnaoaqqsMtMvvx9y9{{}%~%ss__ՑՓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS'SL'SSTS!S!S/%S/%S;:S(;:S~S~SSS)S()SSSSSSS S  SS SSL SgCS gCS+RS +RSg[S g[SSS!ŅS!ŅSSSSDSSS/lS /lS S S߉S0߉SwS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  b> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  b> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  b>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ko&8\c>Y}@d e"Fm(qOs b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@d0T D4$xhX|Hl bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1X , T( } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  LL)6*644>"?"IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbLL 6 6      iiS S""$=%='')'*',,./11335p6p88:Z;Z==?D@DBBD.E.GGIJLLNOQyQySSUcVcXXZM[M]]_7`7bbd!e!ggi j llnnpdqdssuHvHxxz,{,}}ff؉؋JJ.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS5˃S5˃SSLSSTS!S!S%S%S:S(:SSSySySS(SSS%S%SSSS SS SSL SOCS OCSCRS CRS[S [SǚSǚSUDžSUDžS-S-S#SD#SSS5lS 5lS/S/S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zT  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 UO$ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zT  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 UO$ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zT   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 UO$  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` -Fjs7(LUy7 .7[{7?_7 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@42 0 .,*(& bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 i DzU } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  QQ)=*=44>)?)IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))wwc c""$O%O'');*;,,.'/'11346688:u;u==?a@aBBDMEMGGI9J9LLN!O!QQS T V}V}XXZe[e]]_M`Mbbd5e5ggijllnoqyqyssuavaxxzI{I}}11uu]] b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS3SL3SSTS!S!S %S %SA:S(A:S~S~SmSmSS(SSSSSSSS S}S }SSL SCCS CCSRS RS[S [S9S9SSSSS5SD5S S SGlS GlSkSkS;S0;SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 K bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 K bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 K  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ko9>b <\=6Z_4Mq9]  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ D&J,P2V8\>bDhJn bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1"  ` ?   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  ZZ)F*F44>2?2IIST^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffZZ N N  BB66**  "#%%'')t*t,,.`/`113L4L668898;;=$>$@@BCEEGGIrJrLLNZOZQQSBTBVVX*Y*[[]^``bbdnenggiVjVlln>o>qqs&t&vvxy{{}}jjނބRRƇƉ::"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS)˃S)˃SSLSSTS!S!S %S %S:S(:SSSmSmSS(SSSSS}S}SS S}S }SSL SCCS CCSRS RS[S [S'S'SsSsSSS5SD5S S SGlS GlSkSkS;S0;SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1mE b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 KK bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1mE b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 KK bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1mE  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 KK  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`@Lp%I ,Pl=a 0Mq}Uy-Q]m b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ .B2VFjZ~n 0 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 v \ B (  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  UU)C*C44>1?1IIST^^hiszsz} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggUU C C  11  r r""$`%`'')N*N,,.@}@}BBDkEkGGIYJYLLNCOCQQS-T-VVXY[[]^`v`vbbd`e`ggiJjJlln4o4qqstvvxy{}{}}}gg܂܄QQƇƉ;;%% b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS-SL-SSTS!S!S %S %SA:S(A:S~S~SmSmSS(SSSSSSSS S}S }SSL SCCS CCSRS RS[[S [[SSSDžSDžSWSWSSDSSS/lS /lS S S߉S0߉SkS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 M bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 M bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 M  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ko+BfmSw(LSw2[9]cCg  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@PtVz\b hntz!  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 } \;y } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  VV)>*>44>&?&IIS T ^|^|hhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&& ~ ~""$j%j'')R*R,,.:/:113"4"668 9 ;~;~==?f@fBBDNENGGI6J6LLNOQQSSUpVpXXZT[T]]_8`8bbdeggijlrlrnnpVqVssu:v:xxz{}}ttXXʉʋ<<  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS/˃S/˃SSLSSTS!S!S %S %S:S(:S S SmSmSS(SSSSSSSS S}S }SSL SCCS CCS[RS [RS[S [SߚSߚSʅSʅSSS#SD#SSS5lS 5lSSS߉S0߉S}S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1oI b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 KK bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1oI b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 KK bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1oI  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 KK  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`2Pt:8\j> DRv D,:^}$H b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ D4XHl\ p 2 "!F!!!6"Z" bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 tZ@& } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  RR)<*<44>&?&IIS T ^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **xxd d""$O%O'')9*9,,.#/#113 4 6688:l;l==?V@VBBD@E@GGI*J*LLNOQQSSUiViXXZO[O]]_5`5bbdeggijltltnnpZqZssu@v@xxz&{&}} ee؉؋KK11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS9SL9SSTS!S!S%S%SG:S(G:S~S~SsSsS S( SSSSS#S#SS SS SSL SICS ICSQS QSZS ZSCSCSeSeS7S7SSDSSS/lS /lS S S߉S0߉SkS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 L bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 L bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 L  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`8{<`6Y}Uy} 8\+OSw:%)M b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@'K5YCgQu_ m{& bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1J -  s V  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  NN)8*844> ? IIST^y^yhhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''ss] ]""$G%G'')1*1,,./11346z6z88:d;d==?L@LBBD2E2GGIJLLNNPqQqSSUWVWXXZ=[=]]_#`#bbd e g|g|iikblbnnpHqHssu.v.xxz{}}mmSSʼnŋ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS#˃S#˃SSLSSTS!S!S%S%S:S(:S~S~SsSsS S( SSSSSwSwSS SS SSL SICS ICSQS QSZS ZS+S+SSSSS5SD5S S SGlS GlSSSSS#S0#SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b: b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 KM bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b: b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 KM bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b:  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 KM  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ko?Lp/S(Lg7(CgrCkNr b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Lpd|, D8\Pth bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1$ t \ D  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  [[)G*G44>1?1IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kk__ S S  GG;;++  "#%y%y'')e*e,,.Q/Q113=4=668)9);;=>@@BBDqEqGGIYJYLLNAOAQQS)T)VVXY[[]]_m`mbbdUeUggi=j=lln%o%qqs t vvxxzi{i}}QQłń99 yy b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS3SL3SSTS!S!S%S%SG:S(G:S~S~SsSsS S( SSSSS#S#SS SS SSL SICS ICSQS QSZS ZS7S7SȅSȅSSS SD SSSlS lS S Sc߉S0c߉S S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 L bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 L bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 L  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` 0qLp|Os6Zf-Q DPt / .:^z b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@4B,P:^HlVzd r ! bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations19 bE } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  NN)4*444>?IISS]k^khhrMsM} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))qqW W""$=%='')#*#,,. / 1|1|335b6b88:H;H==?,@,BBDEGGIIKaLaNNPCQCSSU%V%XXZ[]x]x__aZbZddf  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 KM bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d>  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 KM bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d>   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 KM  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ko7Z~)M3Vz+ /vRv +TxNr b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ 0$H<`Txl !4!!!("L""" bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 ,|dL } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  UU)=*=44>#?#IIST^y^yhhr]s]} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiYY I I  99))q q""$Y%Y'')A*A,,.)/)11346688:m;m==?S@SBBD7E7GGIJLLNNPqQqSSUUVUXXZ9[9]]_`bbdegsgsiikWlWnnp;q;ssuvxxz{}u}uYY˄ˆ== b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃS9SL9SSTS!S!S%S%SM:S(M:S~S~SySySS(SSSũSũS)S)SS SS SSL SOCS OCSQS QS[S [S[S[So…So…SSS5SD5S S SGlS GlS;S;S S0 SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 L bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 L bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 L  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`7|:^~ 2`PtwDh"FIm(L? b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$$HHll""FFjj  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations18 &   p  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  SS)=*=44>(?(IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&&r r""$\%\'')F*F,,.0/011346689;y;y==?c@cBBDMEMGGI8J8LLN O QQSTV|V|XXZd[d]]_L`Lbbd4e4ggijllnoqxqxssu`v`xxzH{H}}00xxbb b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqS˃S˃SSLSSTS!S!S%S%S:S(:S~S~SsSsS S( SSS S SqSqSS SS SSL SICS ICSQS QSZS ZSSSÅSÅSESESMSDMS%S%S_lS _lSSSwS0wSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W/ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 KT bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W/ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 KT bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W/  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 KT  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ko:@d)M8Sw}, %IwOs Uy!E b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ DNr |7Aeo*4Xb bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1   x k ^  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  ]])O*O44>A?AIIS-T-^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kk]] O O  AA33%%  " # %%'')t*t,,.f/f113X4X668J9J;;=<><@@B.C.EEG H JJLMOOQQSuTuVVXcYc[[]Q^Q``b=c=eeg'h'jjlmooqqsjtjvvxPyP{{}6~6ttXXʑʓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃS9SL9SSTS!S!S%S%SG:S(G:S~S~SySySS(SSSSS#S#SS SS SSL SOCS OCSQS QS%[S %[SmSmS˅S˅S}S}S)SD)SSS;lS ;lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 L bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 L bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 L  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`5}X|+eLpMq@du5Y4Xi b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ DDhhBB f f!!!"""#@### bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 ^L:( } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  QQ);*;44>%?%IIS T ^|^|hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **vv` `""$J%J'')4*4,,./11346}6}88:g;g==?Q@QBBD;E;GGI%J%LLN O Q~Q~SSUdVdXXZJ[J]]_0`0bbdeggiikjljnnpNqNssu2v2xxz{}}llބކRRʼnŋ88 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqS#˃S#˃SSLSSTS!S!S%S%S:S(:S~S~SsSsS S( SSS S SqSqSS SS SSL SICS ICSRS RSO[S O[SSSIͅSIͅSSS5SD5S S SGlS GlS5S5SS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y3 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 KU bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y3 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 KU bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y3  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 KU  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Ko8\/S8\78\8i\ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@'KUy > H!l!!"v"" #1###;$_$$$i%% bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1QD7* } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  WW)E*E44>3?3IIST^^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiWW E E  33!!t t""$b%b'')P*P,,.>/>113,4,6689;;=>@@BBDmEmGGI[J[LLNEOEQQS/T/VVXY[[]^`x`xbbd^e^ggiDjDlln(o(qqs t v{v{xxz]{]}}??!!uuWW b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃSWSWSS$Se:Se:S~S~SaS$aSݩSݩSASASS$SQSQSZS4ZSçST çSk S k SSS SS+S +SelS elS}S }SMSMSS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Me b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Me b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Me b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 pNr,P .y  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ww  ee)S*S44>A?AIIS/T/^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1wwee S S  AA//   ""$p%p'')^*^,,.L/L113:4:668(9(;;=>@@BCE{E{GGIiJiLLNWOWQQSETEVVX3Y3[[]!^!``bceeggitjtlln]o]qqsAtAvvx%y%{{} ~ {{__чщCC'' b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS˃S˃S{S {SSS:S:S~S~SCSCSSpSkSkSSSQS`QSZSZSçS$ çSk S k S_S _S7S 7SqlS qlSS SSS=S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K# b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1c} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K# b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1c} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K# b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1c k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ko)M+v Tx  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  dd)T*T44>B?BIIS.T.^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ttdd T T  DD44$$  "#%|%|'')l*l,,.\/\113L4L668<9<;;=,>,@@BCEEG H JJLLNnOnQQSZTZVVXFYF[[]2^2``bceeg h jjllnlolqqsXtXvvxDyD{{}0~0~~jj b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃSQSQSS$S_:S_:S~S~SaS$aSשSשS;S;SS$SQSQSZS4ZSçST çSk S k SAS ASS SSlS SlS#S #S߉S߉SS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Na b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Na b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Na b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 sQu/S 1|  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10uu  __)I*I44>3?3IIST^^hissss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1uu__ I I  33||f f""$P%P''):*:,,.$/$11346688:m;m==?W@WBBDAEAGGI+J+LLNOQQSSUtVtXXZ^[^]]_H`Hbbd2e2ggijllnoququssuYvYxxz={=}}!!ww[[͎͐?? b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS˃S˃SS SSS:S:S~S~SCSCSSpSkSkSSSQS`QSZSZSçS$ çSk S k SGS GSS SYlS YlSMS MSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K% b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1b} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K% b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1b} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K% b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1b k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ko)M+v Tx  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  ``)L*L44>6?6IIST^^hiszsz} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1tt`` L L  88$$r r""$^%^'')J*J,,.6/6113"4"6689;;==?p@pBBD\E\GGIFJFLLN.O.QQSTVVXXZr[r]]_Z`ZbbdBeBggi*j*llnoqqssunvnxxzV{V}}>>&&jj b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃSQS QSSS_:S_:S~S~SaSaSשSpשS;S;SSSQS`QSZSZSS$ Se S e SMS MS%S %S_lS _lSS SSSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 oMq+O -x  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ww  kk)W*W44>A?AIIS)T)^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1wwkk _ _  SSGG;;//  "###%%'(**,,.q/q113]4]668I9I;;=5>5@@B!C!EEG H JJLLNiOiQQSQTQVVX9Y9[[]!^!``b c e}e}ggiejellnMoMqqs5t5vvxy{{}~yyaaՇՉII11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS˃S˃SS SSS:S:S~S~S1S1S Sp SqSqSSSQS`QSZSZSS$ Se S e SGS GSS SYlS YlSS SSSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T, b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1T} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T, b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1T} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T, b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1T k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ko)M+v Tx  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  ^^)L*L44>8?8IIS"T"^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp^^ L L  ::(( { {""$i%i'')W*W,,.E/E113343668!9!;;=>@@BBDtEtGGI`J`LLNJOJQQS4T4VVXY[[]^`}`}bbdgegggiQjQlln;o;qqs%t%vvxy{{}}nnXX͇͉BB,, b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃSKS KSSSY:SY:S~S~SaSaSѩSpѩS5S5SSSQS`QSZSZSS$ Se S e S5S 5S S  SGlS GlS S  S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 tRv0T2}  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10uu  ee)M*M44>3?3IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1uuee U U  EE55%%  "#%y%y'')a*a,,.I/I1131416689;;=>@u@uBBD]E]GGICJCLLN'O'QQS T V}V}XXZa[a]]_E`Ebbd)e)ggi j llnnpcqcssuGvGxxz+{+}}ee׉׋II-- b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS˃S˃SS SSS:S:S~S~S1S1S Sp SqSqSSSQS`QSZSZSS$ Se S e S5S 5S S  SGlS GlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T. b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1U} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T. b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1U} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T. b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1U k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ko)M+v Tx  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  ]])G*G44>/?/IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qq]] I I  55!!  o o""$Z%Z'')D*D,,../.11346689;w;w==?a@aBBDKEKGGI3J3LLNOQQSSUrVrXXZX[X]]_>`>bbd$e$ggi j l}l}nnpcqcssuIvIxxz/{/}}nnTTǎǐ:: b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃSES ESSSY:SY:S~S~SaSaSѩSpѩS5S5SSSQS`QS+[S+[SS$ S_ S _ S/S /SS SAlS AlS S  S߉S߉SwS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ko8\6Z4X  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  [[)E*E44>+?+IIST^^hhrhsh} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp\\ H H  44    j j""$T%T'')>*>,,.(/(11346688:q;q==?W@WBBD=E=GGI#J#LLN O Q|Q|SSUbVbXXZH[H]]_.`.bbdeggiikmlmnnpSqSssu9v9xxz{}}xx\\Ή΋@@$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS#˃S#˃SS SSS:S:S~S~S1S1SSpSwSwSSSCRS`CRS[S[SS$ S_ S _ S;S ;SS SMlS MlSSS SS#S#SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]5 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1X} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]5 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1X} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]5 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1X k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  0 .,*  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  ``)L*L44>4?4IIST^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ppdd X X  LL@@00  "#%~%~'')j*j,,.V/V113B4B668.9.;;=>@@BCEtEtGGI\J\LLNDODQQS,T,VVXY[[]]_p`pbbdXeXggi@j@lln(o(qqstvvxxzl{l}}TTȂȄ<<!!zz b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃS?S ?SSSS:SS:S~S~SgSgS˩Sp˩S/S/SSSQS`QS7[S7[SS$ S_ S _ SS SS S)lS )lS S  Si߉Si߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 KoBf@d>b  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  YY)?*?44>!?!IIST^t^thhrTsT} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp\\ H H  44    ||b b""$H%H'').*.,,./11335m6m88:S;S==?5@5BBDEGGIIKjLjNNPLQLSSU.V.XXZ[]]__acbcddfEgEiik'l'nnp q szszuuw\x\zz|>}> ppPP b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS#˃S#˃SS SSS:S:S~S~S1S1SSpSwSwSSSRS`RS[S[SS$ S_ S _ S#S #SS S5lS 5lS S  S߉S߉SYS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]7 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Y} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]7 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Y} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]7 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Y k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 "F6ZJn^  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  \\)D*D44>(?(IIS T ^~^~hhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp`` P P  @@00 x x""$`%`'')H*H,,.0/011346689;t;t==?X@XBBD[>]]_"`"bbdegxgxiik\l\nnp@q@ssu$v$xxz{}z}z^^ЄІ@@"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS?S ?SSSS:SS:S~S~SgSgS˩Sp˩S/S/SSSQS`QS7[S7[SS$ S_ S _ S5S 5S S  SGlS GlSAS ASSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 KoDhBf@d  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10oo  \\)F*F44>/?/IIST^^hhrsss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1oo__ O O  ??// { {""$e%e'')O*O,,.9/9113#4#668 9 ;;==?l@lBBDVEVGGI?J?LLN'O'QQSTVVXXZk[k]]_S`Sbbd;e;ggi#j#lln o qqssugvgxxzO{O}}77}}gg b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS)˃S)˃SS SSS:S:S~S~S+S+SSpSwSwSSSCRS`CRS[S[SS$ S_ S _ SAS ASS SSlS SlSS SqSqSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f>  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1T} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f>  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1T} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f>  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1T k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 53 1 /  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  ^^)P*P44>@?@IIS,T,^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ll^^ P P  BB44&&  " # %%'')u*u,,.g/g113Y4Y668K9K;;==>=@@B/C/EEG!H!JJLMOOQQStTtVVXbYb[[]P^P``b*?*IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mmYY E E  11}}g g""$Q%Q'');*;,,.%/%11346688:n;n==?X@XBBDBEBGGI*J*LLNOQQSSUiViXXZO[O]]_5`5bbdeggiikolonnpSqSssu7v7xxz{}}qqUUȉȋ;;!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS)˃S)˃SS SSS:S:SSS+S+SSpS}S}SSS?SS`?SS{\S{\SS$ S_ S _ S/S /SS SAlS AlS/S /S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f@  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1U} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f@  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1U} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f@  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1U k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Cg#Gc'  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  ZZ)H*H44>4?4IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1llZZ H H  66$$ w w""$e%e'')S*S,,.A/A113/4/6689;;= > @@BBDpEpGGI\J\LLNFOFQQS0T0VVXY[[]^`y`ybbd_e_ggiEjElln)o)qqs t v|v|xxz^{^}}@@""ttVV b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS9S 9SSSM:SM:S~S~SaSaSũSpũS)S)SSSQS`QS+[S+[SS$ SY S Y S5S 5S S  SGlS GlS_S _S/S/SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ko<`:^8\  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  YY)G*G44>5?5IIS#T#^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kkYY G G  55##v v""$d%d'')R*R,,.@/@113.4.6689;;= > @@BBDoEoGGI]J]LLNKOKQQS9T9VVX'Y'[[]^``bcezezggihjhllnQoQqqs5t5vvxy{{}}ooSSŇʼn77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS/˃S/˃SS SSS:S:SSS+S+SSpS}S}SSS7RS`7RSs[Ss[SS$ SY S Y SAS ASS SSlS SlSS SwSwSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1oG b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1V} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1oG b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1V} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1oG b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1V k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  .,*(  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  XX)H*H44>6?6IIS"T"^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhXX H H  88((  ""$p%p'')`*`,,.P/P113@4@668090;;= > @@BCEEGHJvJvLLNbObQQSNTNVVX:Y:[[]&^&``bceeggitjtlln`o`qqsLtLvvx8y8{{}$~$rr^^ԑԓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS3S 3SSSG:SG:S~S~SaSaSSpS#S#SSSSS`SSW\SW\SS$ SY S Y S#S #SS S5lS 5lSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 KoCg;_  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  SS)=*=44>'?'IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiSS = =  ''ppZ Z""$D%D'').*.,,./11346w6w88:a;a==?K@KBBD5E5GGIJLLN O Q~Q~SSUhVhXXZR[R]]_<`*?*IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,zzf f""$R%R'')>*>,,.*/*11346689;x;x==?d@dBBDPEPGGI:J:LLN"O"QQS T V~V~XXZf[f]]_N`Nbbd6e6ggijllnoqzqzssubvbxxzJ{J}}22vv^^ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS3S 3SSSG:SG:S~S~S[S[SSpS#S#SSSQS`QS[S[SS$ SS S S S/S /SS SAlS AlSkS kS;S;SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ko.R,P*N  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  __)K*K44>5?5IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kk__ S S  GG;;//##  "#%%'(*y*y,,.e/e113Q4Q668=9=;;=)>)@@BCEEGHJuJuLLN]O]QQSETEVVX-Y-[[]^``bbdqeqggiYjYllnAoAqqs)t)vvxy{{}}mmUUɇɉ==%% b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS/˃S/˃SS SSS:S:S S SSSSpSSSSS1RS`1RSm[Sm[SS$ SS S S S/S /SS SAlS AlSeS eS5S5SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xP b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1P} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xP b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1P} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xP b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1P k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 )'%#  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  RR)@*@44>,?,IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddRR @ @  ..  o o""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGITJTLLN>O>QQS(T(VVXY[[]]_q`qbbd[e[ggiEjElln/o/qqstvvxy{x{x}}bbׂׄLL66  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS-S -SSSA:SA:S~S~S[S[SSpSSSSS]SS`]SS\S\SS$ SS S S SS SS S)lS )lS S  S߉S߉SkS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ko!a7w)M  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  YY)A*A44>'?'IIS T ^}^}hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiYY I I  99))   ""$m%m'')U*U,,.=/=113%4%668 9 ;;==?i@iBBDQEQGGI7J7LLNOQQSSUqVqXXZU[U]]_9`9bbdeggijlslsnnpWqWssu;v;xxz{}}uuYYˉˋ==!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS5˃S5˃SS SSS:S:S S SSSSpSSSSSSS`SS\S\SS$ SS S S SS SS S/lS /lS S  S߉S߉SwS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xR b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Q} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xR b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Q} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xR b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Q k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 [ /k?{  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  QQ);*;44>#?#IIS T ^|^|hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))wwc c""$N%N'')8*8,,."/"113 4 6688:k;k==?U@UBBD?E?GGI'J'LLN O QQSSUfVfXXZL[L]]_2`2bbdeggiikqlqnnpWqWssu=v=xxz#{#}} ||bbՉՋHH.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃS-S -SSS;:S;:S~S~S[S[SSpSSSSSRS`RS \S \SS$ SS S S SS SS S#lS #lS S  S߉S߉S_S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 *NRv z2  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  OO)9*944>?IIST^z^zhhr^s^} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((tt^ ^""$H%H'')2*2,,./11346{6{88:e;e==?K@KBBD1E1GGIJLLNNPpQpSSUVVVXXZ<[<]]_"`"bbd e g}g}iikclcnnpIqIssu/v/xxz{}}nnRRĉċ66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S5˃S5˃SS SSS:S:SSSSS%Sp%SSSSSRS`RS[S[SS$ SS S S S)S )SS S;lS ;lSAS ASSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y% b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1S} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y% b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1S} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y% b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1S k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 m" DBfd  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  TT)@*@44>(?(IIST^^histst} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddXX L L  @@44$$  ""$r%r'')^*^,,.J/J113646668"9";;= > @@BBDhEhGGIPJPLLN8O8QQS T VVXY[|[|]]_d`dbbdVeVggi>j>lln&o&qqstvvxxzj{j}}RRƂƄ::xx b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃS'S 'SSS5:S5:S~S~SUSUSSpSSSSSQS`QS[S[SS$ SS S S SS SS SlS lS S  SQ߉SQ߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 weSwAe  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  MM)3*344>?IISS]l^lhhrLsL} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((ppV V""$<%<'')"*",,./1{1{335a6a88:G;G==?)@)BBD E G|G|IIK^L^NNP@Q@SSU"V"XXZ[]u]u__a[b[ddf=g=iiklnnpqsrsruuwTxTzz|6}6hh؋؍HH b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S;˃S;˃SS SSS:S:SSSSS%Sp%SSSSSQS`QSZSZSS$ SS S S SS SS S#lS #lS S  S߉S߉SAS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[' b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1R} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[' b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1R} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[' b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1R k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Lp*N,w Uy  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP)8*844>?IIST^z^zhhr\s\} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$l l""$T%T'')<*<,,.$/$113 4 6688:h;h==?L@LBBD0E0GGIJLLNNPjQjSSUNVNXXZ2[2]]_`bbdegtgtiikXlXnnp4XNr  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  PP)<*<44>%?%IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33##o o""$[%['')E*E,,.///11346689;x;x==?b@bBBDLELGGI5J5LLNOQQSTVyVyXXZa[a]]_I`Ibbd=e=ggi%j%lln o qqssuivixxzQ{Q}}99!! ii b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17L_S;˃S;˃SS SSS:S:SSSSS+Sp+SSSSSQS`QS%[S%[SS SM S| M S;S ;Sc~S c~SANS ANSS SމSމS9S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b. b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1M} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b. b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1M} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b. b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1M k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1f D4X  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  RR)N*N44>>?>IIS*T*^^hiss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``RR D D  66((   ""$%'')s*s,,.e/e113W4W668I9I;;=;>;@@B-C-EEGHJJL M OOQQSrTrVVX`Y`[[]N^N``b:c:eeg$h$jjlmooqrtstsvvxYyY{{}?~?%% {{__ёѓ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17L_SʃSʃS!S !SSS/:S/:S~S~SOSOSSpS S SSS}QS`}QSZSZSS SM S| M SS S?~S ?~SNS NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1q%)tRv  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  LL):*:44>"?"IIST^^hhrisi} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%qq[ [""$I%I'')3*3,,./11346|6|88:f;f==?P@PBBD:E:GGI"J"LLNOQ{Q{SSUaVaXXZG[G]]_-`-bbdeggijlslsnnpWqWssu;v;xxz{}}uuYỶ̋??%% b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17L_S;˃S;˃SS SSS:S:SSSSS+Sp+SSSSSaRS`aRS[S[SS SM S| M S)S )SQ~S Q~S/NS /NS#S #SމSމSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d0 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1N} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d0 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1N} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d0 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1N k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1KoKoJn7[  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  NN)D*D44>0?0IIST^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``NN < <  **}}k k""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIXJXLLNBOBQQS,T,VVXY[[]^`u`ubbd[e[ggiAjAlln1o1qqstvvxxzf{f}}HH** ||^^ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations1SSʃSʃSS|SSS S S"S|"S&SH&S9S9S]~SP]~SSXSSSSPSSSqSqSS$ S{S, {SS SS$ SS SWDS| WDSQSXQS7[S|7[SSSÅSxÅSS\SUSUSS&SS&S;S8;SSSMlSMlSySySSlSeSeS S } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1= bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1=  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1= bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` 0qJn"SwPt5Y'Vz; -\ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@,Mqe/SC|56ZfMqHl+d*NB{ 0}_Ae# G   ) v  X |  : ^   @   "oQu3W9hJn,P2aCg%I+x Z~<`B b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dMAzg34lNh5O  i  6   P   j7 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10`` RR)N*N44>@?@IIS,T,^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``RR D D qq   ""$%'')s*s,,.e/e113W4W668I9I;;=;>;@@B-C-EEGHJJLMOOQQStTtVVXbYb[[]P^P``b3?3IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeYY M M  AA55))  "#%%'')u*u,,.a/a1134668 9 ;;=%>%@@BCEEGGIsJsLLN[O[QQSCTCVVX+Y+[[]^``bbdoeoggiWjWllnKoKqqs3t3vvxy{{}~ww__ӇӉGG//  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1SSʃSʃSS|SSS S S"S|"S&SH&S9S9S]~SP]~SSXSSSSPSSSqSqSS$ S{S, {SS SS$ SS SWDS| WDSRSXRS\S|\S]S]S)̅Sx)̅SS\SSS(S(S#S8#SSS5lS5lSySyS#Sl#S߉S߉SS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1= bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1=  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1= bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` ..R}.R8c\Imm/SSw!E b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@o&(LNrt+-Q5YSwLp *yc 0  B  { 2 V 5 Y   X | L p   ~  5  7[]:<`^53WUy w,*NLpn#!ECge  !!!"<""":#^###\$$ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1do^M<%+` T xgV  C  0    w dQ>+ bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10`` NN)D*D44>2?2IIST^^hiszsz} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``NN < < pp^^}}k k""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIZJZLLNDODQQS.T.VVXY[[]^`w`wbbd]e]ggiCjClln3o3qqstvvxxzh{h}}JJ,,bb# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xSSY˃SY˃SSdSSlS!S!S$S$S:S@:S/S/SaSaSS@SwS wSCSCSSSS SqS qSSd S7CS 7CSRS RS[S [S-S-SÅSÅSSSUSUSS&S S&SSSSpS)lSt)lSySTyS S S߉S$߉SeS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 M bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 M bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 M  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`CNrr,2VKo:4Xw 1l`9] b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Mq/S5dFj(L .{>b(]g?c-Q! E   V z  '  @ t    V z E i  8 \ /   >   mOs1U7fHl*N 0}_Ae#G)v X|:^@ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dMg4INGh5EO  i  6   P   j7 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  SS)*44>'?'IIS T ^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33## { {""$k%k'')S*S,,.;/;1~1~335f6f88:N;N==?g@gBBDOEOGGI7J7LLNOQQSSUqVqXXZU[U]]_9`9bbdeggijlslsnnpcqcssuGvGxxz+{+}}ee׉׋II--  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSʃSʃS!Sd!SSlS!S!S$S$S5:S@5:S~S~SaSaSS@SwS wSSSSSS SqS qSSd S7CS 7CSRS RS[S [S S SÅSÅSSS[S[SY&S Y&S5S5S Sp SGlStGlSySTySSSkS$kSS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 M bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 M bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 M  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`5Pt;+Rv/S,Tx .GkVz_ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ /Mqj/S(L5 Ei'dbFj D(L=a .{]?c! E   ' t  V z  8 \   >   mOs1U7fHl*N 0}_Ae#G)v X|:^@ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d MFg<4wNh5O  i  6   P   j7 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  PP)@*@44>0?0IvIvST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``PP @ @  00   x x""$h%h'')X*X,,.H/H113848668(9(;;=>@@BCEEGGIpJpLLN\O\QQSHTHVVXY[z[z]]_f`fb c eegginjnllndodqqsPtPvvxp{h8`uXPH@80 * $    zt bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  MM)7*744>?IIST^J^Jhhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&rr\ \""$F%F'')0*0,,./11346y6y88:c;c==?K@KBBD1E1GGIJLLNNPpQpSSUVVVXXZ<[<]]_"`"bbd e g}g}iikclcnnpqssuuwrxrz{}}nnTTƉƋ88  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSʃSʃS!Sd!SSlS!S!S$S$S/:S@/:S~S~SaSaSS@SwS wSSS S SS SqS qSSd S7CS 7CSQS QS[S [SISISÅSÅSSS[S[SY&S Y&SSSSpS/lSt/lSySTyS#S#S߉S$߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 M bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 M bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 M  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` /t/S| \5^Dh@dd,P"FFj b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Mq/S5dFj(L .{Z~<]w?c5Y! E   R v  '  4 t o  V z / S  8 \   >   mOs1U7fHl*N 0}_Ae#G)v X|:^@ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dMg42Nm(hc5O  i  6   P   j7 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  LL)8*844>$?$IgIgS T ^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``LL 8 8  $$rr^ ^""$J%J'')6*6,,."/"11346688:p;p==?\@\BBDHEHGGI4J4LLNOQQSTVGVGXXZ/[/]]_`bbd0e0ggijllnoq|q|ssudvdxxzL{L}}44xx``  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSM˃SM˃SSdSSlS!S!S$S$S:S@:S)S)SaSaSS@SwS wS=S=SSSS SqS qSSd S7CS 7CScSS cSS\S \S盅S盅S/ɅS/ɅSSSSS'S 'SSSSpS lSt lSySTyS S SQ߉S$Q߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yE b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 M^ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yE b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 M^ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yE  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 M^  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`"FtPt 0Rv<`Ei>l(Jn:^ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@cq*8"F0T5r>b4XLpW{Z ~  = z   h < `   " v _  !   /  ='K5YAeKoUy_is}& 0: D*N4X>bHlR v  \!! b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dcF) o2oR,5j'{^?    b  C  $fG( bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  KK)1*144>?IISS];^;hhrNsN} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&nnT T""$:%:'') * ,,./1y1y335_6_88:E;E==?)@)BBD E G|G|IIK^L^NNP@Q@SSU"V"XXZ[]u]u__a[b[ddf=g=iikln_n_pprAsAuuw#x#zz|6}6jjڋڍJJ  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃS'SL'SSTS!S!S%S%S5:S(5:S~S~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CSRS RS[S [S9S9SSSSS#SD#SSS5lS 5lSYSYS)S0)SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 M bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 M bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 M  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`:Vz D4TxDh 1Rv /hPt b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@(e)MLp2o1UTx bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1A~;x } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  NN)<*<44>*?*IIST^^hhrsss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``NN < <  **}}k k""$Y%Y'')G*G,,.5/5113#4#6689;;==?v@vBBDdEdGGIRJRLLNbg:49]y< /Kot>A@ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$kAe;X|.R(oEi bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1GU } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  RR)<*<44>'?'IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%%q q""$[%['')E*E,,.///11346689;x;x==?b@bBBDLELGGI7J7LLNOQQSTV{V{XXZc[c]]_K`Kbbd3e3ggijllnoqwqwssu_v_xxzG{G}}//wwaa b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃS'SL'SSTS!S!S%S%S5:S(5:S~S~SgSgSS(S}S}SSSSSS SwS wSSL S=CS =CSQS QSZS ZS=S=SSSSSSDSSS#lS #lS S S߉S0߉SkS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 M bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 M bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 M  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` 0s0T{Uy2Y}}7[7[[?99] b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@:w9]\ B   A e  ' d  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations15r/l) } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  MM)7*744>!?!IIST^z^zhhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%ss_ _""$J%J'')4*4,,./11346}6}88:g;g==?Q@QBBD;E;GGI%J%LLN O Q~Q~SSUdVdXXZJ[J]]_0`0bbdeggiikolonnpUqUssu;v;xxz!{!}}zz``ӉӋFF,, b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SG˃SG˃SSLSSTS!S!S %S %S:S(:S#S#SmSmSS(SSS7S7SSSS S}S }SSL SCCS CCSRS RS\S \SWSWSŴSŴSSSSDSSS#lS #lS S S߉S0߉SYS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n: b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 MT bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n: b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 MT bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n:  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 MT  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`%Iv .4Y}6Dh>'Kx"F6[ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@?\2 V  , s  I m  C   bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1c*q8 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  NN)8*844>"?"IIST^y^yhhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''ss] ]""$G%G'')1*1,,./11346z6z88:d;d==?N@NBBD8E8GGI"J"LLNOQ{Q{SSUaVaXXZG[G]]_-`-bbdeggiikglgnnpKqKssu/v/xxz{}}iiۄۆOO‰‹55 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃS-SL-SSTS!S!S%S%S;:S(;:S~S~SsSsS S( SSSSSSSS SS SSL SICS ICS RS  RSI[S I[SSSSSSS#SD#SSS5lS 5lSASASS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  O  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  O  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  O   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Mq?\Bf?\DhA^FjC` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$h8\,p@d4xHl< bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1DL } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  TT)@*@44>*?*IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddXX L L  @@44$$  ""$r%r'')^*^,,.J/J113646668"9";;=>@@BBDjEjGGIRJRLLN:O:QQS"T"VVX Y [~[~]]_f`fbbdNeNggi6j6llnoqqstvzvzxxzb{b}}JJ22rr b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S;˃S;˃SSLSSTS!S!S%S%S:S(:SSSSSS(SSS+S+SSSS SS S SL  SUCS UCScSS cSS\S \S盅S盅SSSSS)SD)SSS;lS ;lSSSSS#S0#SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_+ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 RT. bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_+ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 RT. bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_+  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 RT.  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`,DhpNr>Fj@[@d1U 2V b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$vbNr:^&J6"t bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1R$vH } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>1?1IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIYJYLLNGOGQQS5T5VVX#Y#[[]^``bbdvevggidjdllnMoMqqs1t1vvxy{{}}kk݂݄OO33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃS'SL'SSTS!S!S %S %S;:S(;:S~S~SmSmSS(SSSSSSSS S}S }SSL SCCS CCS+RS +RSg[S g[SSSkSkSSS SD SSSlS lS S S߉S0߉SAS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 N bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 N bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 N  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`` 11Uz h /Txx p.RRv,x-Q  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ Pt D X | ( L  ` 0 T  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1T\ d } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP)8*844>?IIST^t^thhrXsX} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$l l""$T%T'')<*<,,.$/$113 4 6688:h;h==?N@NBBD2E2GGIJLLNNPlQlSSUPVPXXZ4[4]]_`bbddfngniikRlRnnp6q6ssuvxxzz|p}pTTƄƆ88 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SA˃SA˃SSLSSTS!S!S%S%S:S(:SSSSSS(SSS1S1SSSS SS S SL  SUCS UCSRS RS!\S !\SiSiSSS9S9SSDSSS)lS )lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c/ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 SS0 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c/ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 SS0 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c/  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 SS0  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`(Lx98`; Lp =4X? Dl b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@`Lp8 \ $ H  4 r ^ Jn bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1l>b } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  MM)7*744>#?#IIS T ^^hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccMM 7 7  !!  jjT T""$>%>'')(*(,,./11335q6q88:[;[==?E@EBBD/E/GGIJLLNOQzQzSSUdVdXXZN[N]]_8`8bbd"e"ggi j llnnpeqessuIvIxxz-{-}}ggىًKK// b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS-SL-SSTS!S!S5%S5%SA:S(A:S~S~SSS/S(/SSSSSSSS SS S%SL %SmCS mCS'SS 'SSc\S c\SSSSSSS;SD;SSSMlS MlSSSeS0eS S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  e@ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  e@ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  e@  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Mq Db ,PnKo8\z$Dh 0Im b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$s Y}?c%I /~dJn bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1Om< } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  VV)H*H44>:?:IIS&T&^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddVV H H  ::,,  "#%{%{'')m*m,,._/_113Q4Q668C9C;;=5>5@@B'C'EEGHJJL M OOQQSnTnVVX\Y\[[]J^J``b6c6eeg h jjl m o}o}qqsctcvvxIyI{{}/~/mmߌߎQQÑÓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS5˃S5˃SSLSSTS!S!S%S%S:S(:SSSySySS(SSS%S%SSSS SS SSL SOCS OCSSS SSW\S W\SSSgSgSOSOS5SD5S S SGlS GlSkSkS;S0;SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zT  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 RI" bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zT  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 RI" bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zT   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 RI"  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` /Jny ?c&JUy&1Uwp7 1LpKo  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$&(*, . 02 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1]:t } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  ]])I*I44>5?5IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ii]] Q Q  EE99--!!  "#%%'(*w*w,,.c/c113O4O668;9;;;='>'@@BCEEGGIuJuLLN]O]QQSETEVVX-Y-[[]^``bbdqeqggiYjYllnAoAqqs)t)vvxy{{}}mmUUɇɉ==%% b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS-SL-SSTS!S!S/%S/%SA:S(A:S~S~SSS)S()SSSSSSS S  SS SSL SgCS gCSRS RS=[S =[SSSǵSǵS!S!S#SD#SSS5lS 5lS#S#S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  d> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  d> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  d>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Mq22VyJn,OssLp%IImNrC b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@0T: o  U y ; _ ! E  + z  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 Z)xG } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  RR)@*@44>.?.IIST^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddRR @ @  ..  o o""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGIVJVLLN@O@QQS*T*VVXY[[]]_s`sbbdYeYggi?j?lln#o#qqstvvvvxxzX{X}}::ppRR b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS;˃S;˃SSLSSTS!S!S%S%S:S(:SSSySySS(SSS+S+SSSS SS SSL SOCS OCSRS RS3\S 3\S{S{SwSwSSSSDSSS/lS /lS S S߉S0߉SkS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|X$ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 SI$ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|X$ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 SI$ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|X$  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 SI$  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`+Oz<<g@)Tx DAe$H.R} b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@4 6  8  :  <  > @B bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1Q. hE } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  WW)?*?44>'?'IIS T ^}^}hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggWW G G  77''  ""$k%k'')S*S,,.;/;113#4#668 9 ;;==?g@gBBDOEOGGI7J7LLNOQQSSUqVqXXZU[U]]_9`9bbdeggijlslsnnpWqWssu;v;xxz{}}uuYYˉˋ==!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS3SL3SSTS!S!S%S%SG:S(G:S~S~SsSsS S( SSSSS#S#SS SS SSL SICS ICSRS RS!\S !\SiSiS=S=SISISASDASSSSlS SlSSSwS0wSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 N bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 N bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 N  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Mq&Ji>b-Q 1Vz%Imn 1=a b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$~ zvrnj fb bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1Z4h } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  TT)D*D44>4?4IIS T ^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$ | |""$l%l'')\*\,,.L/L113<4<668,9,;;=>@@B C EEGGItJtLLN`O`QQSLTLVVX8Y8[[]$^$``bceeggirjrlln^o^qqsJtJvvx6y6{{}"~"pp\\ґғ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS/˃S/˃SSLSSTS!S!S%S%S:S(:S S SsSsS S( SSSSSSSS SS SSL SICS ICSQS QSZS ZS1S1SSSSSSDSSS/lS /lS S S߉S0߉SkS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1oI b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 MM bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1oI b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 MM bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1oI  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 MM  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`2MqPt/S8\j5 DRv,e:^| b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$<0THl`x(@4X bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1hP8  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  QQ);*;44>#?#IIS T ^|^|hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **vv` `""$J%J'')4*4,,./11346}6}88:g;g==?O@OBBD5E5GGIJLLNOQtQtSSUZVZXXZ@[@]]_&`&bbd e ggiikelennpKqKssu1v1xxz{}}ppVVȉȋ:: b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS3SL3SSTS!S!S%S%SG:S(G:S~S~SsSsS S( SSSSS#S#SS SS SSL SICS ICSIRS IRS[S [S͚S͚SSSSS)SD)SSS;lS ;lS/S/S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 N bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 N bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 N  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Mq33Wx)aJnn@u@d#67 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@^ Z ~ V z R v N r J n FjBf bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1BvP* } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP)<*<44>(?(IIST^^hhrlsl} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((vvb b""$N%N''):*:,,.&/&11346688:t;t==?`@`BBDLELGGI8J8LLN O QQSTV|V|XXZd[d]]_L`Lbbd4e4ggijllnoqxqxssu`v`xxzH{H}}00tt\\ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS5˃S5˃SSLSSTS!S!S%S%S:S(:S S SsSsS S( SSSSSSSS SS SSL SICS ICSSS SSK\S K\SSS͸S͸SSS SD SSSlS lS S S]߉S0]߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1qM b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 MM bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1qM b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 MM bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1qM  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 MM  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`.RMq|/S@n52\ dNr<` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@L p d  |  , D 8\Pth bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1pX@( } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  OO)5*544>?IISS]l^lhhrNsN} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **rrX X""$>%>'')$*$,,. / 1}1}335c6c88:I;I==?-@-BBDEGGIIKbLbNNPDQDSSU&V&XXZ[]y]y__a[b[ddf=g=iiklnnpqsrsruuwTxTzz|6}6jjڋڍJJ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS9SL9SSTS!S!S%S%SM:S(M:S~S~SySySS(SSSũSũS)S)SS SS SSL SOCS OCSRS RS[S [S'S'S寅S寅SySyS;SD;SSSMlS MlSqSqSAS0ASS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 N bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 N bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 N  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Mq*Nn$qNr'$Hr'K(HlKo(Ll b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@%7%I7[Im[m, bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1fK0 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  RR)@*@44>.?.IIST^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddRR @ @  ..  o o""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGIVJVLLN@O@QQS*T*VVXY[[]]_s`sbbd]e]ggiGjGlln1o1qqstvvxy{z{z}}ddقلNNÇÉ88"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS)˃S)˃SSLSSTS!S!S%S%S:S(:SSSySySS(SSSSS}S}SS SS SSL SOCS OCSQS QS[S [SgSgS3S3SSS5SD5S S SGlS GlS;S;S S0 SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d>  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 MK bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d>  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 MK bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d>   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 MK  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Mq5Vz5Y)Jn D>bs,3W|h b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$@8\Txp">6ZRv bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1jT>( } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  VV)@*@44>+?+IIST^^hhrqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiYY I I  99)) u u""$_%_'')I*I,,.3/311346689;|;|==?f@fBBDPEPGGI;J;LLN#O#QQS T VVXXZg[g]]_O`Obbd7e7ggijllnoq{q{ssucvcxxzK{K}}33{{ee b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS9SL9SSTS!S!S%S%SM:S(M:S~S~SySySS(SSSũSũS)S)SS SS SSL SOCS OCS[RS [RS[S [SߚSߚSySySSS#SD#SSS5lS 5lS S S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 N bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 N bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 N  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Mq45Yw'eEij7}8\z -+!E b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ > , P > b P t b t!3 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1z_D) } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  QQ);*;44>%?%IIS T ^~^~hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))wwc c""$N%N'')8*8,,."/"113 4 6688:k;k==?U@UBBD?E?GGI)J)LLNOQQSSUhVhXXZN[N]]_4`4bbdeggijlslsnnpYqYssu?v?xxz%{%}} ~~dd׉׋JJ00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS/˃S/˃SSLSSTS!S!S%S%S:S(:SSSySySS(SSSSS}S}SS SS SSL SOCS OCS!SS !SS]\S ]\SSSSSSS#SD#SSS5lS 5lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1fB b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 MJ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1fB b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 MJ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1fB  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 MJ  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Mq1U~3W Du;;d*m[!Jn b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ n  < 4 X P t l:2V bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1|fP: } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  RR)<*<44>&?&IIS T ^}^}hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++wwa a""$K%K'')5*5,,./113 4 6~6~88:h;h==?R@RBBD.?.IIST^^hhrrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hh\\ P P  DD88((  "#%v%v'')b*b,,.N/N113:4:668&9&;;=>@@BBDnEnGGIVJVLLN>O>QQS&T&VVXY[[]]_j`jbbdReRggi:j:lln"o"qqs t v~v~xxzf{f}}NN‚„66vv b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqS#˃S#˃SSLSSTS!S!S%S%S:S(:S~S~SySySS(SSS S SqSqSS SS SSL SOCS OCSQS QSZS ZS1S1S۰S۰SSSASDASSSSlS SlSkSkS;S0;SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y3 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 MR bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y3 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 MR bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y3  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 MR  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Mq 1Nr~/S5Ye5?Ko%d1Ut b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$2Vd!/Sa,P^  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1uj_T } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  WW)E*E44>5?5IIS#T#^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiWW E E  33!!t t""$b%b'')P*P,,.>/>113,4,6689;;=>@@BBDmEmGGI]J]LLNKOKQQS9T9VVX'Y'[[]^``bcezezggihjhllnQoQqqs5t5vvxy{{}}ooSSŇʼn77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃS?SL?SSTS!S!S%S%SM:S(M:S~S~SySySS(SSSũSũS)S)SS SS SSL SOCS OCSRS RSU[S U[SSSϹSϹSYSYS#SD#SSS5lS 5lS S S߉S0߉SSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 N bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 N bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 N  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`59]v 'v@dh .j2Vo  _ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@ d h $ H L p t,0TX| bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations10 p } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  TT)<*<44>"?"IIST^x^xhhr\s\} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhXX H H  88((p p""$X%X'')@*@,,.(/(11346688:l;l==?R@RBBD6E6GGIJLLNNPpQpSSUTVTXXZ8[8]]_`bbdegrgriikVlVnnp:q:ssuvxxz{}t}tXXʄʆ<< b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqS#˃S#˃SSLSSTS!S!S%S%S:S(:S~S~SySySS(SSSSSwSwSS SS SSL SOCS OCSRS RS\S \SWSWSkSkSSS/SD/SSSAlS AlSSS߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[7 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 MS bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[7 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 MS bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[7  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 MS  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Mq,Pt 06Z\B<`*$Hzl(Tx b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ ) M [   & J X|#GUy bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1I>3(  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  QQ);*;44>'?'IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggQQ ; ;  %%nnX X""$B%B''),*,,,./11346u6u88:_;_==?I@IBBD3E3GGIJLLN O Q~Q~SSUhVhXXZR[R]]_<`M?MIIS9T9^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1yykk ] ]  OOAA33%%  "#%%' ( **,,.t/t113f4f668X9X;;=J>J@@B@?@IIS(T(^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvjj ^ ^  RRFF::..  ""#"%%'(**,,.p/p113\4\668H9H;;=4>4@@B C EEG H JJLLNhOhQQSPTPVVX8Y8[[] ^ ``bce|e|ggidjdllnLoLqqs4t4vvxy{{}~xx``ԇԉHH00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃSWSWSS$Se:Se:S~S~SaS$aSݩSݩSASASS$SQSQS[S4[SçST çSk S k SGS GSS SYlS YlSGS GSSSS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Jd b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Jd b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Jd b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 z\>b"F  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ww  ee)S*S44>???IIS'T'^^h i sysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1wwee S S  AA//   ""$p%p'')^*^,,.L/L113:4:668(9(;;=>@@BCE{E{GGIgJgLLNQOQQQS;T;VVX%Y%[[]^``bbdjejggiPjPlln4o4qqstvvxxzi{i}}KK--aa b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS˃S˃SS SSS:S:S~S~SCSCS Sp SqSqSSSQS`QSZSZSçS$ çSk S k S;S ;SS SMlS MlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M) b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1`} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M) b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1`} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M) b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1` k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Mq/S7f  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10vv  ff)N*N44>4?4IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvff V V  FF66&&  "#%z%z'')b*b,,.J/J1132426689;;=>@v@vBBD^E^GGIDJDLLN(O(QQS T V~V~XXZb[b]]_F`Fbbd*e*ggijllnnpdqdssuHvHxxz,{,}}ff؉؋JJ.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃSQS QSSSe:Se:S~S~SaSaSݩSpݩSASASSSQS`QS[S[SS$ Se S e SYS YS1S 1SklS klSS SSS7S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 qSw5Y;  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10uu  ee)U*U44>C?CIIS/T/^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1uuee U U  EE55%%  "#%}%}'')m*m,,.]/]113M4M668=9=;;=->-@@BCEEG H JJLLNoOoQQS[T[VVXGYG[[]3^3``bceeg h jjllnmomqqsYtYvvxEyE{{}1~1 kk b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS#˃S#˃SS SSS:S:S~S~S1S1SSpSwSwSSSaRS`aRS[S[SS$ Se S e S/S /SS SAlS AlSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V0 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1R} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V0 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1R} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V0 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1R k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 :<>@  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10oo  ZZ)D*D44>*?*IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1oo[[ G G  33  i i""$S%S'')=*=,,.'/'11346688:p;p==?V@VBBD5?5IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ss__ K K  77##q q""$]%]'')I*I,,.5/5113!4!668 9 ;;==?o@oBBD[E[GGIEJELLN-O-QQSTVVXXZq[q]]_Y`YbbdAeAggi)j)llnoqqssumvmxxzU{U}}==%% ii b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS#˃S#˃SS SSS:S:S~S~S1S1SSpSwSwSSSRS`RS[S[SS$ Se S e SS SS S/lS /lS S  So߉So߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V2 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1S} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V2 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1S} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V2 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1S k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 B*N6ZBf  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  ZZ)@*@44>"?"IIST^u^uhhrUsU} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qq]] I I  55!!  }}c c""$I%I'')/*/,,./11335n6n88:T;T==?6@6BBDEGGIIKkLkNNPMQMSSU/V/XXZ[]]__adbdddfFgFiik(l(nnp q s{s{uuw]x]zz|?}?!!qqQQ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃSKS KSSS_:S_:S~S~SgSgSשSpשS;S;SSSQS`QS[S[SS$ S_ S _ SGS GSS SYlS YlS}S }SMSMSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 qSw5Y;  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  __)M*M44>9?9IIS#T#^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qq__ M M  ;;)) | |""$j%j'')X*X,,.F/F113444668"9";;=>@@BBDuEuGGIaJaLLNKOKQQS5T5VVXY[[] ^ `~`~bbdhehggiRjRlln.?.IIST^^hhrrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nn^^ N N  >>.. z z""$d%d'')N*N,,.8/8113"4"668 9 ;;==?k@kBBDUEUGGI>J>LLN&O&QQSTVVXXZj[j]]_R`Rbbd:e:ggi"j"lln o q~q~ssufvfxxzN{N}}66||ff b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃSES ESSSY:SY:S~S~SgSgSѩSpѩS5S5SSSQS`QS [S [SS$ S_ S _ S5S 5S S  SGlS GlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 z\>b D  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  \\)F*F44>.?.IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp\\ H H  44    n n""$Y%Y'')C*C,,.-/-11346689;v;v==?`@`BBDJEJGGI2J2LLNOQQSSUqVqXXZW[W]]_=`=bbd#e#ggi j l|l|nnpbqbssuHvHxxz.{.}}mmSSƎƐ99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS)˃S)˃SS SSS:S:SSS+S+SSpS}S}SSSESS`ESS\S\SS$ S_ S _ S)S )SS S;lS ;lS S  S߉S߉SwS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_; b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Q} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_; b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Q} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_; b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Q k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ei %Ie)  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  YY)C*C44>+?+IIST^^hhrfsf} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nnZZ F F  22  ~~h h""$R%R'')<*<,,.&/&11346688:o;o==?Y@YBBDCECGGI+J+LLNOQQSSUjVjXXZP[P]]_6`6bbdeggiikplpnnpTqTssu8v8xxz{}}rrVVɉɋ<<"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSES ESSSS:SS:S~S~SaSaS˩Sp˩S/S/SSSQS`QS1[S1[SS$ S_ S _ S;S ;SS SMlS MlSSS SS#S#SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Mq8\:^<`  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  aa)M*M44>5?5IIST^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qqee Y Y  MMAA11  " # %%'')k*k,,.W/W113C4C668/9/;;=>@@BCEuEuGGI]J]LLNEOEQQS-T-VVXY[[]]_q`qbbdYeYggiAjAlln)o)qqstvvxxzm{m}}UUɂɄ==""{{ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS/˃S/˃SS SSS:S:SSS+S+SSpS}S}SSSURS`URS[S[SS$ S_ S _ S;S ;SS SMlS MlSeS eS5S5SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1hB b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1V} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1hB b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1V} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1hB b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1V k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 8:<>  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  XX)F*F44>4?4IIS"T"^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjXX F F  44""u u""$c%c'')Q*Q,,.?/?113-4-6689;;= > @@BBDnEnGGI\J\LLNJOJQQS8T8VVX&Y&[[]^``bceyeyggigjgllnPoPqqs4t4vvxy{{}}nnRRćĉ66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS?S ?SSSS:SS:S~S~SaSaS˩Sp˩S/S/SSS%RS`%RSa[Sa[SS$ S_ S _ S#S #SS S5lS 5lS S  S߉S߉SSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 MqUy]e   b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10oo  [[)C*C44>'?'IIS T ^}^}hhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1oo__ O O  ??// w w""$_%_'')G*G,,.///11346688:s;s==?W@WBBD;E;GGIJLLNOQuQuSSUYVYXXZ=[=]]_!`!bbdegwgwiik[l[nnp?q?ssu#v#xxz{}y}y]]τφ??!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS/˃S/˃SS SSS:S:S S S1S1SSpSSSSSSS`SS\S\SS$ S_ S _ S)S )SS S;lS ;lS S  S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1hD b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1W} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1hD b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1W} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1hD b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1W k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Z~,d6n  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  TT)>*>44>(?(IIST^^hhrhsh} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjTT > >  ((qq[ [""$E%E'')/*/,,./11346x6x88:b;b==?L@LBBD6E6GGI J LLN O QQSSUiViXXZS[S]]_=`=bbd'e'ggijllnnpjqjssuNvNxxz2{2}}llމދPPŽ44 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS?S ?SSSM:SM:S~S~SgSgSũSpũS)S)SSSRS`RS=[S=[SS$ SY S Y SAS ASS SSlS SlSS SkSkSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 MqGkImKo  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  __)Q*Q44>A?AIIS-T-^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mm__ Q Q  CC55''  " # %%'')v*v,,.h/h113Z4Z668L9L;;=>>>@@B0C0EEG"H"JJLMOOQQSuTuVVXcYc[[]Q^Q``b=c=eeg'h'jjlmooqqsjtjvvxPyP{{}6~6rrVVȑȓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS/˃S/˃SS SSS:S:S S S%S%SSpSSSSSIRS`IRS[S[SS$ SY S Y S5S 5S S  SGlS GlSkS kS;S;SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1qK b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1P} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1qK b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1P} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1qK b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1P k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  1357  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  ^^)J*J44>4?4IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jj^^ R R  FF::..""  "#%%'(*x*x,,.d/d113P4P668<9<;;=(>(@@BCEEGHJtJtLLN\O\QQSDTDVVX,Y,[[]^``bbdpepggiXjXlln@o@qqs(t(vvxy{{}}llTTȇȉ<<$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS9S 9SSSM:SM:S~S~SaSaSũSpũS)S)SSSRS`RS[S[SS$ SY S Y S)S )SS S;lS ;lS)S )S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Mq ?Os   b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  YY)G*G44>3?3IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kkYY G G  55##v v""$d%d'')R*R,,.@/@113.4.6689;;= > @@BBDoEoGGI[J[LLNEOEQQS/T/VVXY[[]^`x`xbbd^e^ggiDjDlln(o(qqs t v{v{xxz]{]}}??!!ssUU b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS5˃S5˃SS SSS:S:S S S%S%SSpSSSSSSS`SS\S\SS$ SY S Y SS SS S/lS /lS S  S߉S߉SqS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1qM b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1O} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1qM b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1O} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1qM b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1O k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ] 1mA}  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  ZZ)B*B44>(?(IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjZZ J J  ::**   ""$n%n'')V*V,,.>/>113&4&6689;;==?j@jBBDRERGGI8J8LLNOQQSTVrVrXXZV[V]]_:`:bbdeggijltltnnpXqXssu>"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS3S 3SSSG:SG:S~S~SUSUSSpS#S#SSSQS`QS1[S1[SS$ SS S S SAS ASS SSlS SlSS SwSwSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Mq>b@dBf  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  YY)I*I44>7?7IIS#T#^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiYY I I  99))   ""$q%q'')a*a,,.Q/Q113A4A668191;;=!>!@@BCEEGHJwJwLLNcOcQQSOTOVVX;Y;[[]'^'``bceeggiujullnaoaqqsMtMvvx9y9{{}%~%ss__ՑՓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS5˃S5˃SS SSS:S:SSSSS%Sp%SSSSSyRS`yRS[S[SS$ SS S S SS SS S)lS )lS S  S߉S߉S_S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zT  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1R} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zT  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1R} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zT  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1R k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 k6,PFj  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  NN)8*844>?IIST^w^whhr[s[} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''ss] ]""$G%G'')1*1,,./11346z6z88:d;d==?J@JBBD0E0GGIJLLNNPoQoSSUUVUXXZ;[;]]_!`!bbdegzgziik`l`nnpFqFssu,v,xxz{}}kkބކOO33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS3S 3SSSG:SG:S~S~SUSUSSpS#S#SSSiSS`iSS\S\SS$ SS S S S)S )SS S;lS ;lS/S /S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Mq%iA9]  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  SS)?*?44>)?)IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++yye e""$Q%Q'')=*=,,.)/)11346689;w;w==?c@cBBDOEOGGI9J9LLN!O!QQS T V}V}XXZe[e]]_M`Mbbd5e5ggijllnoqyqyssuavaxxzI{I}}11uu]] b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS;˃S;˃SS SSS:S:SSS%S%S%Sp%SSSSSQS`QSZSZSS$ SS S S SS SS SlS lS S  SW߉SW߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zV" b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1S} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zV" b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1S} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zV" b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1S k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Nr2V:l  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  NN)4*444>?IISS]i^ihhrIsI} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))qqW W""$=%='')#*#,,. / 1|1|335b6b88:H;H==?*@*BBD E G}G}IIK_L_NNPAQASSU#V#XXZ[]v]v__aXbXddf:g:iiklnnpprosouuwQxQzz|3}3eeՋՍEE b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃS-S -SSSA:SA:S~S~S[S[SSpSSSSSQS`QS[S[SS$ SS S S S/S /SS SAlS AlSeS eS5S5SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Mq2V4X6Z  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>-?-IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIUJULLN?O?QQS)T)VVXY[[]]_r`rbbd\e\ggiFjFlln2o2qqstvvxy{{{{}}eeڂڄOOćĉ99## b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S;˃S;˃SS SSS:S:SSSSS+Sp+SSSSSIRS`IRS[S[SS$ SS S S S#S #SS S5lS 5lS/S /S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]) b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1M} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]) b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1M} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]) b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1M k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 dt! 1A  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  OO)9*944>"?"IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbRR B B  22""n n""$X%X'')B*B,,.,/,11346689;u;u==?_@_BBDIEIGGI2J2LLNOQQSTVvVvXXZ^[^]]_F`Fbbd8e8ggi j llnoq|q|ssudvdxxzL{L}}44zzdd b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃS-S -SSS;:S;:S~S~S[S[SSpSSSSSSS`SS\S\SS$ SS S S SS SS S/lS /lS S  S߉S߉SwS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Mq+{5Yc  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP):*:44>"?"IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((vvb b""$M%M'')7*7,,.!/!113 4 6688:j;j==?T@TBBD>E>GGI&J&LLN O QQSSUeVeXXZK[K]]_1`1bbdeggiikplpnnpZqZssu@v@xxz&{&}} ee؉؋KK11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_S;˃S;˃SS SSS:S:SSSSS+Sp+SSSSSQS`QSZSZSS$ SS S S SS SS S)lS )lS S  S߉S߉SeS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_+ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1L} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_+ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1L} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_+ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1L k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 MqPt2V8  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  MM)7*744>?IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&rr\ \""$F%F'')0*0,,./11346y6y88:c;c==?M@MBBD7E7GGIJLLNOQxQxSSU^V^XXZD[D]]_*`*bbdeggiiklllnnpPqPssu4v4xxz{}}nnRRʼnŋ88 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃS'S 'SSS;:S;:S~S~SISISSpSSSSSCRS`CRS[S[SS SM S| M S)S )SQ~S Q~S/NS /NSGS GS=މS=މSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1,PX|>  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  UU)C*C44>+?+IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeYY M M  AA55%%  ""$u%u'')a*a,,.M/M113949668%9%;;=>@@BBDkEkGGISJSLLN;O;QQS#T#VVX Y [[]]_g`gbbd[e[ggiCjClln+o+qqstvvxxzo{o}}WW˂˄??$$ }} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17L_SA˃SA˃SS SSS:S:SSSSS1Sp1SSSSSQS`QSZSZSS SM S| M S5S 5S]~S ]~S;NS ;NS_S _SUމSUމSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f2 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1O} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f2 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1O} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f2 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1O k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1]}"$  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10^^  LL)D*D44>2?2IIS T ^^hiss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1^^LL : :  (({{i i""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIZJZLLNHOHQQS6T6VVX$Y$[[]^``bcewewggiejellnZoZqqs>t>vvx"y"{{}~xx\\·Ή@@$$ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17L_SʃSʃS'S 'SSS5:S5:S~S~SOSOSSpSSSSSeQS`eQSZSZSS SM S| M SS S?~S ?~SNS NS S  S݉S݉SmS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1waKo8\  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  OO);*;44>?IIST^^hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33##k k""$W%W'')?*?,,.'/'11346688:k;k==?O@OBBD3E3GGIJLLNNPmQmSSUQVQXXZ5[5]]_`bbd e g{g{iik_l_nnpCqCssu'v'xxz { }}}}aaӄӆCC%% b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17L_SA˃SA˃SS SSS:S:SSSSS1Sp1SSSSS-SS`-SSi\Si\SS SM S| M S#S #SK~S K~S)NS )NSS S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h4 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1P} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h4 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1P} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h4 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1P k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1Mq ."F(  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10^^  HH):*:44>$?$IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1^^HH 2 2  {{eeO O""$A%A'')+*+,,./11335t6t88:^;^==?H@HBBD2E2GGIJLLNOQ{Q{SSUeVeXXZO[O]]_9`9bbd#e#ggi j llnoqrqrssuVvVxxz:{:}}ttXXʎʐ<< b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations1YSʃSʃSS|SSS S S"S|"S&SH&S9S9S]~SP]~SSXSSSSPSSSqSqSS$ S{S, {SS SS$ SS SWDS| WDSQSXQSZS|ZSCSCS'ŅSx'ŅSS\SSS&S&S)S8)SSS;lS;lSySyS5Sl5SSSS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1? bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1?  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1? bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$t'Kl %Z~Dhn@d@Hl&J"F b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Qu;_%I3nX|Bf,P: $ u   _  I m  3 W  (  a  A  ?  x + |2VjIm'X|`Fj>4X"F4"ucQu?c-Q? -n \ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dQ"sDf7YD*}}6 P o # (  v  I   o Bh bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__ LL)@*@44>+?+IIST^^his}s}} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__OO ? ? uuee  k k""$_%_'')I*I,,.3/311346689;|;|==?f@fBBDPEPGGI;J;LLN#O#QQS T VVXXZg[g]]_O`ObbdCeCggi+j+llnoqqssuovoxxzW{W}}??''qq# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSY˃SY˃SSdSSlS!S!S$S$S:S@:S/S/SaSaSS@SwS wSCSCSSSS SqS qSSd S7CS 7CSRS RS[S [S-S-SmąSmąS)S)SSS}&S }&S/S/SSpSAlStAlSySTySeSeS5S$5SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 O bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 O bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 O  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Os>h 2VA 0Jn$HQub<` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Os5Y?%t Z~@d&J 0eK o  1 U   ;   ! p   V z  <`~"Fd,{Jna0TGk:.R o8Uym;_Sw9]C)x^Dh*N b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dOm< Z)xGe24   P   S n  " q @ ^- bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  PP)*44>.?.IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$]%]'')K*K,,.9/91~1~335l6l88:Z;Z=>@z@zBBDhEhGGIVJVLLN@O@QQS*T*VVXY[[]]_s`sbbd]e]ggiGjGlln=o=qqs't'vvxy{{}}ppZZχωDD..  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1YSʃSʃSS|SSS S S"S|"S&SH&S9S9S]~SP]~SSXSSSSPSSSqSqSS$ S{S, {SS SS$ SS SWDS| WDS3SSX3SSo\S|o\SSS΅Sx΅SS\SSS)S)SS8SSS)lS)lSySyS Sl S߉S߉SeS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1? bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1?  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1? bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`5LpDh /Jn )Hl*N%Fjn4 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@z<Tx.Rj,Dh B  Z ~    4 X  p 2Jn $Hw`1U"Hl8\&_p=v 00TDh|<Pt$H\0Th !(!!!<"`"""t##$4$$$H%l%%&&&'@' b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dztnhb\VPJD < 4 a  , S $  E |t bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__ JJ)<*<44>&?&IIST^^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__KK 7 7 z zffRRooY Y""$K%K'')5*5,,./113 4 6~6~88:h;h==?R@RBBDPt@6Hl& .u@d b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1 bWuoperation.4ipu_stack_copy_u64%:%R)2))2)")B)Z-b1reasonLocalCopyvalue8bytes144patternContiguousinvocations16- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1Os5Y?%t Z~@d&J2gM q  3 W   =   # r  X |  >b$H .}cIm/S9n$ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1<Om< Z)xGe4   R  ! & bWuoperation.6ipu_stack_copy_u64%:%R)2))2)")B)Z-b1reasonLocalCopyvalue10bytes512patternContiguousinvocations5+ bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  MM)*44>%?%IIS T ^^hhrpsp}- b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1. b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1/ bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%ss_ _""$N%N'')8*8,,."/"1d1d335N6N88:8;8==?U@UBBD?E?GGI)J)LLNOQQSSUhVhXXZN[N]]_4`4bbdeggijlslsnnpeqessuKvKxxz1{1}}ppVVɎɐ<<5 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations16 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSʃSʃS'Sd'SSlS!S!S%S%S5:S@5:S~S~SgSgSS@S}S }SSSSSS SwS wSSd S=CS =CSRS RS=[S =[SSSmąSmąS)S)SSS}&S }&S)S)SSpS;lSt;lSySTySSSSS#S$#SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 O bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 O bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 O  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$*Ni",x;_w 2lUyy  `3W b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Os5Y?%t Z~@d&J 0eK o  1 U   ;   !  p J n    , X | g  >b%I$HBf .$}_Ac|Im/S9nTx:^ D*y_Ei+O b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dOm< Z)xGWeM 4   E  R  ! p ? ], bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  MM);*;44>+?+IpIpST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__MM ; ;  ))||j j""$X%X'')F*F,,.4/4113"4"6689;;==?u@uBBDcEcGGISJSLLNAOAQQS/T/VtVtXXZb[b]]_P`Pbbdpepggi^j^llnQoQqqs5t5vvxy{{}}ooSSŇʼn77  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSS˃SS˃SSdSSlS!S!S%S%S:S@:S)S)SgSgSS@S}S }S=S=SSSS SwS wSSd S=CS =CSRS RS[S [SSS/̅S/̅SSSSS(S (S#S#SSpS5lSt5lSySTyS;S;S S$ SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yE b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Oa bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yE b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Oa bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yE  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Oa  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`An -$Ae4$He:Hl@?l b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@m" DBfd;9][ }  2  0 T  R v   t )/S'K5ImdkFj (L"F .{Hl]n%'KMqs*,PRv x /1UW {  !}!!"4"""6#Z###\$$ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dmZG4!{hUB ^  1 +  x   E ~ m \K:) bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  VV)B*B44>,?,IIST^X^Xhhrrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffZZ N N  BB66&&  ""$t%t'')`*`,,.L/L113848668$9$;;=>@@BBDlElGGITJTLLN?IaIaS T ^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__II 3 3  ||ffP P""$:%:'')$*$,,./11335m6m88:W;W==?A@ABBD+E+GGIJLLNOQvQvSSU-V-XXZ[]]_`bbdeggijl}l}nnpiqissuMvMxxz1{1}}kk݉݋OO33  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSS˃SS˃SSdSSlS!S!S%S%S:S@:S/S/SgSgSS@S}S }SCSCSSSS SwS wSSd S=CS =CS SS  SSE\S E\SSSqƅSqƅS-S-S!S!S'S 'S S SSpSlStlSySTyS S S߉S$߉S;S(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}I b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 O` bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}I b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 O` bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}I  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 O`  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`!EVz8\;Lp /pAe 0$5YzDh b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@X|PtHl@d8\0T(L D< 4   ,   $ |   t   l  d\PtDh8\,P D8?c -!E!'wt kVz _8\Sw> Gkm;_/S#G b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dX0`8h@pF   r H    t _  J , yvFL bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP)8*844>?IIST^E^Ehhr\s\} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$l l""$T%T'')<*<,,.$/$113 4 6688:h;h==?N@NBBD2E2GGIJLLNNPlQlSSUPVPXXZ4[4]]_`bbdegrgriikVlVnnpqsysyuuw]x]z{}t}tXXʄʆ<<  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃS-SL-SSTS!S!S %S %S;:S(;:S~S~SmSmSS(SSSSSSSS S}S }SSL SCCS CCSRS RS!\S !\SiSiSSS-S-S#SD#SSS5lS 5lS_S_S/S0/SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 O bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 O bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 O  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$/Sf+O55Y[;dFj#G b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$ H G k -j,POs5r bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1d!^  X  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  WW)C*C44>/?/IIST^^hhrsss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccWW K K  ??33''  "#%%'')q*q,,.]/]113I4I668595;;=!>!@@B C EEGGIoJoLLNWOWQQS?T?VVX'Y'[[]^``bbdkekggiSjSlln;o;qqs#t#vvx y {{}}ggۂۄOOÇÉ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SG˃SG˃SSLSSTS!S!S %S %S:S(:S#S#SmSmSS(SSS7S7SSSS S}S }SSL SCCS CCS SS  SSE\S E\SSS3S3SSS5SD5S S SGlS GlSSS_S0_SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n: b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 OM" bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n: b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 OM" bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n:  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 OM"  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` Dt 0?cPt <`s 1<l(AeHl b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ Qu/S 1|Z~8\:c bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1p;  Q   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  XX)J*J44>7@@B)C)EEGHJJL M OOQQSpTpVVX^Y^[[]L^L``b8c8eeg"h"jjl m ooqqsetevvxKyK{{}1~1ooSSőœ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃS'SL'SSTS!S!S %S %S;:S(;:S~S~SmSmSS(SSSSSSSS S}S }SSL SCCS CCSRS RSC[S C[SSSSSSSSDSSS#lS #lS S S߉S0߉S_S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 O bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 O bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 O  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`?xX|A_Z~EiC\-Q!E^ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@4XW{=z<`"_!E bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 R  L  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  SS);*;44>#?#IIST^y^yhhr]s]} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33## { {""$g%g'')O*O,,.7/711346689;{;{==?c@cBBDKEKGGI3J3LLNOQQSSUmVmXXZQ[Q]]_5`5bbdeggiikolonnpSqSssu7v7xxz{}}qqUUljNj99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SM˃SM˃SSLSSTS!S!S%S%S:S(:S)S)SsSsS S( SSS=S=SSSS SS SSL SICS ICSCRS CRS[S [SǚSǚS#ÅS#ÅSSSSDSSS/lS /lSSS߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1r> b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ON# bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1r> b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ON# bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1r>  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ON#  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` DTx+4Dh 1$4Xx7$Hh= b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@AeC!lJn(L*uSw bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 g 2 } H } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  RR)@*@44>.?.IIST^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddRR @ @  ..  o o""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGIVJVLLN@O@QQS*T*VVXY[[]]_s`sbbdYeYggi?j?lln#o#qqstvvvvxxzX{X}}::ppRR b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS-SL-SSTS!S!S%S%SA:S(A:S~S~SsSsS S( SSSSSSSS SS SSL SICS ICSQS QS1[S 1[SySyS S SSSSDSSS#lS #lS S S߉S0߉S_S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Q  bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Q  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Q   bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Os&6Ze5Y@@d:^&&Ju @d 0[ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ $ j >b6| Pt$Hb6Z bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1(n4 z @  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  KK)5*544>?IIST^v^vhhr\s\} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``LL 8 8  $$ppZ Z""$D%D'').*.,,./11346w6w88:a;a==?I@IBBD/E/GGIJLLNNPnQnSSUTVTXXZ:[:]]_ ` bbdegygyiik_l_nnpEqEssu+v+xxz{}}jj݄݆PP‰‹44 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSA˃SA˃SSLSSTS!S!S%S%S:S(:SSSSSS(SSS1S1SSSS SS S SL  SUCS UCSuSS uSS\S \SSSSSSS;SD;SSSMlS MlSSSqS0qSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c/ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 PV/ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c/ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 PV/ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c/  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 PV/  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`#Gv4SwW{DhKo%Ix6Y}Cg b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@d0T D4$xhX|Hl bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1\0  X ,  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  VV)F*F44>6?6IIS"T"^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&& ~ ~""$n%n'')^*^,,.N/N113>4>668.9.;;=>@@BCEEGGIvJvLLNbObQQSNTNVVX:Y:[[]&^&``bceeggitjtlln`o`qqsLtLvvx8y8{{}$~$rr^^ԑԓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS-SL-SSTS!S!S%S%SA:S(A:S~S~SsSsS S( SSSSSSSS SS SSL SICS ICSRS RS[S [S S SSSwSwSSDSSSlS lS S SQ߉S0Q߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 P bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 P bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 P  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`k D^&Jd< ,Pj6Z2VpTx b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ .tJnB\0T(n bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1  L  X } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  KK)1*144>?IISS]h^hhhrJsJ} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&nnT T""$:%:'') * ,,./1y1y335_6_88:E;E==?)@)BBD E G|G|IIK^L^NNP@Q@SSU"V"XXZ[]u]u__aWbWddf9g9iiklnnpprnsnuuwPxPzz|2}2ff֋֍FF b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSG˃SG˃SSLSSTS!S!S%S%S:S(:S#S#SSSS(SSS7S7SSSS SS S SL  SUCS UCSsRS sRS[S [SSSŅSŅSSS#SD#SSS5lS 5lS)S)S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g3 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 QU0 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g3 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 QU0 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g3  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 QU0  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`CRv D />b,P*Nm8\:Y}Dh b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@8\(L<,p `Pt bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 T (|P } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP)<*<44>(?(IIST^^hhrlsl} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((vvb b""$N%N''):*:,,.&/&11346688:t;t==?`@`BBDLELGGI8J8LLN O QQSTV|V|XXZd[d]]_L`Lbbd4e4ggijllnoqxqxssu`v`xxzH{H}}00tt\\ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS3SL3SSTS!S!S5%S5%SG:S(G:S~S~SSS/S(/SSSSS#S#SS SS S%SL %SmCS mCSRS RS\S \SKSKSSS5S5S#SD#SSS5lS 5lS/S/S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  g@ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  g@ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  g@  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Os)>bj(LQu#G6;_ [%Iq b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@`Jn4XB,}gQu;_ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1g 8 Z  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  PP):*:44>%?%IIS T ^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33##o o""$Y%Y'')C*C,,.-/-11346689;v;v==?`@`BBDJEJGGI5J5LLNOQQSTVyVyXXZa[a]]_I`Ibbd1e1ggijllnoququssu]v]xxzE{E}}--uu__ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS;˃S;˃SSLSSTS!S!S%S%S:S(:SSSySySS(SSS+S+SSSS SS SSL SOCS OCS-SS -SSi\S i\SSS#S#S%S%S5SD5S S SGlS GlSkSkS;S0;SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|X$ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 PG" bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|X$ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 PG" bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|X$  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 PG"  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`&Jx8Dh^Lp B2V DZ~j b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@#G)M/S5Y;_AeGkMq bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1J )  g F  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  TT)B*B44>0?0IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffTT B B  00  q q""$_%_'')M*M,,.;/;113)4)6689;;=>@|@|BBDjEjGGIXJXLLNBOBQQS,T,VVXY[[]^`u`ubbd_e_ggiIjIlln3o3qqstvvxy{|{|}}ffۂۄPPŇʼn::$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS3SL3SSTS!S!S/%S/%SG:S(G:S~S~SSS)S()SSSSS#S#S S  SS SSL SgCS gCSIRS IRS[S [S͚S͚SaąSaąSISISSDSSS)lS )lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  f> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  f> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  f>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Os&Je=a6Zu!BfFj 1HlVzA b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@%I3nX|Bf,P: bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1+ | Mo } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  NN)8*844>"?"IIST^y^yhhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''ss] ]""$G%G'')1*1,,./11346z6z88:d;d==?N@NBBD8E8GGI"J"LLNOQ{Q{SSUaVaXXZG[G]]_-`-bbdeggiikglgnnpKqKssu/v/xxz{}}iiۄۆOO‰‹55 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSA˃SA˃SSLSSTS!S!S%S%S:S(:SSSySySS(SSS+S+SSSS SS SSL SOCS OCSgRS gRS[S [S뚅S뚅SWȅSWȅS{S{SSDSSS/lS /lS S S߉S0߉SwS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~\( b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 QG$ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~\( b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 QG$ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~\(  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 QG$  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`BPtA*8\z%I Db-Q,Jn|5Y b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@SwY}_e kqw }  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1% cB! } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  QQ);*;44>%?%IIS T ^~^~hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))wwc c""$N%N'')8*8,,."/"113 4 6688:k;k==?U@UBBD?E?GGI)J)LLNOQQSSUhVhXXZN[N]]_4`4bbdeggijlslsnnpYqYssu?v?xxz%{%}} ~~dd׉׋JJ00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS9SL9SSTS!S!S%S%SM:S(M:S~S~SySySS(SSSũSũS)S)SS SS SSL SOCS OCSSS SSK\S K\SSS'S'SSS/SD/SSSAlS AlS_S_S/S0/SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 P bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 P bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 P  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Os .Hlv6ZdAe$HRv 6@d3W b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@d@d@d@dBfDhFjHl bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 ` <  t  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  QQ)?*?44>/?/IIST^^hissss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccQQ ? ?  --  n n""$\%\'')J*J,,.8/8113&4&6689;;=>@y@yBBDgEgGGIWJWLLNEOEQQS3T3VVX!Y![[]^``bbdtetggibjbllnKoKqqs/t/vvxy{{}}iiۂۄMM11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS5˃S5˃SSLSSTS!S!S%S%S:S(:SSSySySS(SSS%S%SSSS SS SSL SOCS OCSRS RS9\S 9\SSS-…S-…SSS5SD5S S SGlS GlSMSMSS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1qM b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 OO bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1qM b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 OO bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1qM  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 OO  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`)MOsz<Gke'?cTx7[?c.R{ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$@8\Txp">6ZRv bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations18 " v `  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  ZZ)F*F44>0?0IIST^^histst} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jj^^ R R  FF::**  "#%x%x'')d*d,,.P/P113<4<668(9(;;=>@@BBDpEpGGIXJXLLN@O@QQS(T(VVXY[[]]_l`lbbdTeTggi#?#IIS T ^^hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccMM 7 7  !!  jjT T""$>%>'')(*(,,./11335q6q88:[;[==?E@EBBD/E/GGIJLLNOQzQzSSUdVdXXZN[N]]_8`8bbd"e"ggi j llnnpeqessuIvIxxz-{-}}ggىًKK// b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS;˃S;˃SSLSSTS!S!S%S%S:S(:SSSySySS(SSS%S%SSSS SS SSL SOCS OCSQS QSZS ZS1S1S˅S˅SSSSSSDSSS/lS /lS S S߉S0߉SMS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1sQ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 OO bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1sQ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 OO bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1sQ  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 OO  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`AOsPt'5Y4Xu ?<Y}% =atn b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@n <4XPt l !!!":"""2#V# bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 J4r } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  TT)<*<44>"?"IIST^x^xhhr\s\} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhXX H H  88((p p""$X%X'')@*@,,.(/(11346688:l;l==?R@RBBD6E6GGIJLLNNPpQpSSUTVTXXZ8[8]]_`bbdegrgriikVlVnnp:q:ssuvxxz{}t}tXXʄʆ<< b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃS?SL?SSTS!S!S%S%SS:S(S:S~S~SSSS(SSS˩S˩S/S/SS SS S SL  SUCS UCSRS RS[S [S9S9S%S%SsSsS;SD;SSSMlS MlSqSqSAS0ASS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 P bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 P bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 P  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Os3Rv "Dhw<6ZiHl(L[x b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@!E7[Mqcy(>0TFj bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1s Z A (   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  [[)G*G44>3?3IIST^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1gg[[ O O  CC77++  "#%%'')u*u,,.a/a113M4M668999;;=%>%@@BCEEGGIsJsLLN[O[QQSCTCVVX+Y+[[]^``bbdoeoggiWjWlln?o?qqs't'vvxy{{}}kk߂߄SSLJlj;;## b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqS/˃S/˃SSLSSTS!S!S%S%S:S(:SSSySySS(SSSSS}S}SS SS SSL SOCS OCSRS RS[S [S9S9S7ąS7ąS]S]SGSDGSSSYlS YlSSSqS0qSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1fB b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 OI bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1fB b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 OI bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1fB  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 OI  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Os&Jr*;_Nr'K.Rv7 .Vz2V~ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@`+O](LZ~%IW{ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 &    z  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  \\)N*N44>@?@IIS,T,^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jj\\ N N  @@22$$  "#%%'')s*s,,.e/e113W4W668I9I;;=;>;@@B-C-EEGHJJLMOOQQStTtVVXbYb[[]P^P``b'?'IIS T ^}^}hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggWW G G  77''  ""$k%k'')S*S,,.;/;113#4#668 9 ;;==?g@gBBDOEOGGI7J7LLNOQQSSUqVqXXZU[U]]_9`9bbdeggijlslsnnpWqWssu;v;xxz{}}uuYYˉˋ==!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqS/˃S/˃SSLSSTS!S!S%S%S:S(:S S SySySS(SSSSSSSS SS SSL SOCS OCSQS QSZS ZS=S=SͅSͅS%S%S/SD/SSSAlS AlS/S/S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1hF b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 OH bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1hF b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 OH bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1hF  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 OH  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Os<Dh7["FeC$Cgo+!E{Mq b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@"FTx !C!!!Q"u""###$@$$$N%r%% & bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1odYNC } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  VV)D*D44>2?2IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhVV D D  22  s s""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIZJZLLNDODQQS.T.VVXY[[]^`w`wbbd]e]ggiCjClln'o'qqs t vzvzxxz\{\}}>> ttVV b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃSESLESSTS!S!S%S%SS:S(S:S~S~SSSS(SSS˩S˩S/S/SS SS S SL  SUCS UCS7RS 7RSs[S s[SSS/ÅS/ÅSSS)SD)SSS;lS ;lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 P bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 P bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 P  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`6X|( .Pt$&Hl~!@d{v b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@:Bfn(0T\BJn bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1` R D 6 (  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  OO)9*944>!?!IIST^z^zhhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((tt^ ^""$H%H'')2*2,,./11346{6{88:e;e==?M@MBBD3E3GGIJLLNNPrQrSSUXVXXXZ>[>]]_$`$bbd e g}g}iikclcnnpIqIssu/v/xxz{}}nnTTƉƋ88 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqS)˃S)˃SSLSSTS!S!S%S%S:S(:S~S~SSSS(SSSSSwSwSS SS S SL  SUCS UCS RS  RSI[S I[SSSAƅSAƅSSSMSDMS%S%S_lS _lSSSS0S+S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[7 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 OQ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[7 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 OQ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[7  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 OQ  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Os?Z~5Y=X|?;Vz%9tTx b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@d,Pt<`(Lp8\ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1     } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  ZZ)J*J44>:?:IIS&T&^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjZZ J J  ::**   ""$r%r'')b*b,,.R/R113B4B668292;;=">"@@BCEEGHJzJzLLNfOfQQSRTRVVX>Y>[[]*^*``bceeghjxjxllndodqqsPtPvvx?IISS]l^lhhrNsN} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **rrX X""$>%>'')$*$,,. / 1}1}335c6c88:I;I==?-@-BBDEGGIIKbLbNNPDQDSSU&V&XXZ[]y]y__a[b[ddf=g=iiklnnpqsrsruuwTxTzz|6}6jjڋڍJJ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqS)˃S)˃SSLSSTS!S!S%S%S:S(:SSSSSS(SSSSS}S}SS SS S SL  SUCS UCSQS QS[S [SOSOSЅSЅSSS5SD5S S SGlS GlS;S;S S0 SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]; b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 OQ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]; b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 OQ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1];  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 OQ  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Os //Sx ;_ -Rvv'K,PPt7,,P b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$H l !!!4"X"""|## $D$$$h%% &0&&&T'x''((( bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  TT)@*@44>,?,IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,zzf f""$R%R'')>*>,,.*/*11346689;x;x==?d@dBBDPEPGGI8?8IIS T ^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1xxhh X X  HH88%%  ""$n%n'')X*X,,.B/B113,4,6689;;=>@u@uBBD_E_GGIHJHLLN0O0QQSTVVXY[t[t]]_\`\bbdDeDggi,j,llnoqqssupvpxxzX{X}}@@((pp b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S#˃S#˃SS SSS:S:S~S~SCSCSSpSwSwSSSQS`QSZSZSçS$ çSk S k SSS SS+S +SelS elSS SYSYSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O+ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1_} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O+ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1_} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O+ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1_ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Os5Y?%t  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  aa)O*O44>;?;IIS%T%^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ssaa O O  ==++ ~ ~""$l%l'')Z*Z,,.H/H113646668$9$;;=>@@BCEwEwGGIcJcLLNMOMQQS7T7VVX!Y![[] ^ ``bbdjejggiTjTlln>o>qqs(t(vvxy{{}}qq[[ЇЉEE// b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}SʃSʃSWSWSS$Sk:Sk:S~S~SgS$gS㩄S㩄SGSGSS$SQSQS1[S41[SçST çSk S k S;S ;SS SMlS MlSS S߉S߉SS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fe b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fe b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Fe b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 iQu9]  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10vv  aa)K*K44>3?3IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvbb N N  ::&&p p""$Z%Z'')D*D,,../.11346689;w;w==?a@aBBDKEKGGI3J3LLNOQQSSUrVrXXZX[X]]_>`>bbd"e"ggijlxlxnnp\q\ssu@v@xxz${$}}zz^^щыDD** b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S#˃S#˃SS SSS:S:S~S~SCSCSSpSwSwSSSQS`QSZSZSçS$ çSk S k SAS ASS SSlS SlS#S #S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O- b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1^} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O- b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1^} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O- b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1^ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Os5Y?'v  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  ``)J*J44>2?2IIST^^hhrqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1tt`` L L  88$$r r""$]%]'')G*G,,.1/111346689;z;z==?d@dBBDNENGGI6J6LLNOQQSTVuVuXXZ[[[]]_A`Abbd'e'ggi j llnnpfqfssuLvLxxz2{2}}qqWWʎʐ== b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃSWS WSSSk:Sk:S~S~SaSaS㩄Sp㩄SGSGSSSQS`QS[S[SS$ Se S e SMS MS%S %S_lS _lSwS wSGSGSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 u [Ae'K  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  bb)P*P44>>?>IIS,T,^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ttbb P P  >>,,  ""$m%m'')[*[,,.I/I113747668%9%;;=>@@BCExExGGIfJfLLNTOTQQSBTBVVX0Y0[[]^``b c eeggiqjqllnZoZqqs>t>vvx"y"{{}~xx\\·Ή@@$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS)˃S)˃SS SSS:S:S~S~S1S1SSpSwSwSSSRS`RS\S\SS$ Se S e SAS ASS SSlS SlS_S _S/S/SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X4 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1P} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X4 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1P} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X4 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1P k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 2VVz z 0  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  cc)O*O44>7?7IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ssgg [ [  OOCC33  " # %%'')m*m,,.Y/Y113E4E668191;;=>@@BCEwEwGGI_J_LLNGOGQQS/T/VVXY[[]]_s`sbbd[e[ggiCjClln+o+qqstvvxxzo{o}}WW˂˄??$$ }} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃSQS QSSSe:Se:S~S~SaSaSݩSpݩSASASSSRS`RS=[S=[SS$ Se S e S;S ;SS SMlS MlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 o [Gk  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10rr  \\)F*F44>0?0IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1rr\\ F F  00yyc c""$M%M'')7*7,,.!/!113 4 6688:j;j==?T@TBBD>E>GGI(J(LLNOQQSSUqVqXXZ[[[]]_E`Ebbd/e/ggijllnoqrqrssuVvVxxz:{:}}ttXXʎʐ<< b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS)˃S)˃SS SSS:S:SSS1S1SSpS}S}SSS-SS`-SSi\Si\SS$ Se S e S/S /SS SAlS AlS S  S߉S߉S_S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X6 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Q} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X6 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Q} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X6 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Q k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Aey9Mq   b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  __)G*G44>+?+IIST^^hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1sscc S S  CC33 { {""$c%c'')K*K,,.3/311346689;w;w==?[@[BBD?E?GGI#J#LLNOQyQySSU]V]XXZA[A]]_%`%bbd e g{g{iik_l_nnpCqCssu'v'xxz { }}}}aaӄӆCC%% b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃSQS QSSS_:S_:S~S~SgSgSשSpשS;S;SSSQS`QS[S[SS$ Se S e SGS GSS SYlS YlS}S }SMSMSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 s Y}?c%I  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  hh)T*T44>>?>IIS&T&^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1tthh \ \  PPDD88,,  " # %%' ( **,,.n/n113Z4Z668F9F;;=2>2@@BCEEG H J~J~LLNfOfQQSNTNVVX6Y6[[]^``bcezezggibjbllnJoJqqs2t2vvxy{{}~vv^^҇҉FF.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS/˃S/˃SS SSS:S:SSS+S+SSpS}S}SSSsRS`sRS[S[SS$ Se S e SMS MS%S %S_lS _lSS SwSwSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a=  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1L} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a=  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1L} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a=  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1L k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 A%I-Q5Y  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10oo  aa)S*S44>C?CIIS/T/^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ooaa S S  EE77))  " # %%'')x*x,,.j/j113\4\668N9N;;=@>@@@B2C2EEG$H$JJLMOOQRTwTwVVXeYe[[]S^S``b?c?eeg)h)jjlmooqqsltlvvxRyR{{}8~8ttXXʑʓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃSKS KSSS_:S_:S~S~SgSgSשSpשS;S;SSSQS`QS+[S+[SS$ Se S e S/S /SS SAlS AlSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 iQu9]  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10rr  bb)J*J44>0?0IIST^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1rrbb R R  BB22""  "#%v%v'')^*^,,.F/F113.4.6689;;==?r@rBBDZEZGGI@J@LLN$O$QQSTVzVzXXZ^[^]]_B`Bbbd&e&ggi j l|l|nnp`q`ssuDvDxxz({(}} ~~bbԉԋFF** b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS/˃S/˃SS SSS:S:S S S%S%SSpSSSSSSS`SS\S\SS$ Se S e S5S 5S S  SGlS GlS5S 5SSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a?  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1K} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a?  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1K} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a?  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1K k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 \ .f8p  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10oo  ]])K*K44>7?7IIST^^hisqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1oo]] K K  99'' z z""$h%h'')V*V,,.D/D113242668 9 ;;=>@@BBDsEsGGI_J_LLNIOIQQS3T3VVXY[[]^`|`|bbdbebggiHjHlln,o,qqstvvxxza{a}}CC%%wwYY b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃSKS KSSSY:SY:S~S~SaSaSѩSpѩS5S5SSSQS`QS7[S7[SS$ S_ S _ S)S )SS S;lS ;lS S  S߉S߉SqS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Os5Y6Z;_  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  XX)B*B44>(?(IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mmYY E E  11}}g g""$Q%Q'');*;,,.%/%11346688:n;n==?T@TBBD:E:GGI J LLNOQyQySSU_V_XXZE[E]]_+`+bbdeggiikjljnnpPqPssu6v6xxz{}}uuYYˉˋ==!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS/˃S/˃SS SSS:S:S S S1S1SSpSSSSSgRS`gRS[S[SS$ S_ S _ SGS GSS SYlS YlSS SSS+S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1jF b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1X} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1jF b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1X} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1jF b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1X k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ;A#G)M  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  [[)K*K44>9?9IIS%T%^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kk[[ K K  ;;++   ""$s%s'')c*c,,.S/S113C4C668393;;=#>#@@BCEEGHJyJyLLNeOeQQSQTQVVX=Y=[[])^)``bceeghjwjwllncocqqsOtOvvx;y;{{}'~'uuaaבד b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃSES ESSSY:SY:S~S~SaSaSѩSpѩS5S5SSSRS`RSO[SO[SS$ S_ S _ SS SS S#lS #lS S  Si߉Si߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 OsHlNrTx  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  VV)<*<44>?IIST^q^qhhrQsQ} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mmYY E E  11yy_ _""$E%E'')+*+,,./11335j6j88:P;P==?2@2BBDEGGIIKgLgNNPIQISSU+V+XXZ [ ]~]~__a`b`ddfBgBiik$l$nnpqswswuuwYxYzz|;};mm݋ݍMM b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS5˃S5˃SS SSS:S:S S S1S1SSpSSSSSSS`SS\S\SS$ S_ S _ S5S 5S S  SGlS GlS;S ;S S SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1jH b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Y} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1jH b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Y} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1jH b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Y k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Y})] -a  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  WW)C*C44>-?-IIST^^hhrqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kkWW C C  //}}i i""$U%U'')A*A,,.-/-11346689;{;{==?g@gBBDSESGGI=J=LLN%O%QQS T VVXXZi[i]]_Q`Qbbd9e9ggi!j!lln o q}q}ssuevexxzM{M}}55yyaa b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS?S ?SSSS:SS:S~S~SgSgS˩Sp˩S/S/SSSRS`RS=[S=[SS$ SY S Y S/S /SS SAlS AlS;S ;S S SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Os<`BfHl  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  YY)C*C44>,?,IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ll\\ L L  <<,, x x""$b%b'')L*L,,.6/6113 4 668 9 ;;==?i@iBBDSESGGI/?/IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggUU C C  11  r r""$`%`'')N*N,,.@}@}BBDkEkGGIWJWLLNAOAQQS+T+VVXY[[]]_t`tbbd^e^ggiHjHlln2o2qqstvvxy{{{{}}eeڂڄOOćĉ99## b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS?S ?SSSM:SM:S~S~SgSgSũSpũS)S)SSS[RS`[RS[S[SS$ SY S Y S#S #SS S5lS 5lS S  S߉S߉SkS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Osg /#G  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  UU)?*?44>'?'IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjVV B B  ..zzd d""$N%N'')8*8,,."/"113 4 6688:k;k==?U@UBBD?E?GGI'J'LLN O QQSSUfVfXXZL[L]]_2`2bbdeggiiklllnnpPqPssu4v4xxz{}}nnRRʼnŋ88 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS;˃S;˃SS SSS:S:SSSSS%Sp%SSSSSSS`SS\S\SS$ SY S Y S#S #SS S5lS 5lSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1sQ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1I} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1sQ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1I} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1sQ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1I k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 \ .f8p  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  TT)>*>44>&?&IIS T ^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,zzf f""$Q%Q'');*;,,.%/%11346688:n;n==?X@XBBDBEBGGI*J*LLNOQQSSUiViXXZO[O]]_5`5bbdeggijltltnnpZqZssu@v@xxz&{&}} ee؉؋KK11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS9S 9SSSM:SM:S~S~SUSUSũSpũS)S)SSSRS`RSC[SC[SS$ SY S Y S/S /SS SAlS AlSYS YS)S)SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 OsFjLpRv  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  VV)D*D44>2?2IIS T ^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhVV D D  22  s s""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIZJZLLNHOHQQS6T6VVX$Y$[[]^``bcewewggiejellnNoNqqs2t2vvxy{{}}llނބPP‡‰44 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS;˃S;˃SS SSS:S:SSS%S%S+Sp+SSSSS1RS`1RSm[Sm[SS$ SY S Y S)S )SS S;lS ;lSAS ASSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|X$ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1T} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|X$ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1T} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|X$ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1T k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 b jrz#  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  WW)C*C44>+?+IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1gg[[ O O  CC77''  ""$u%u'')a*a,,.M/M113949668%9%;;=>@@BBDkEkGGISJSLLN;O;QQS#T#VVX Y [[]]_g`gbbdOeOggi7j7llnoqqstv{v{xxzc{c}}KK33qq b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS9S 9SSSG:SG:S~S~SUSUSSpS#S#SSSSS`SSK\SK\SS$ SY S Y SS SS S/lS /lS S  S߉S߉SwS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 OsAe 3W  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  PP):*:44>$?$IIST^^hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffPP : :  $$mmW W""$A%A'')+*+,,./11335t6t88:^;^==?H@HBBD2E2GGIJLLNOQ{Q{SSUeVeXXZO[O]]_9`9bbd#e#ggi j llnnpfqfssuJvJxxz.{.}}hhډڋLL00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeS;˃S;˃SS SSS:S:SSS%S%S+Sp+SSSSSQS`QSZSZSS$ SY S Y SS SS S#lS #lS S  S߉S߉SGS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|Z& b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1U} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|Z& b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1U} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|Z& b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1U k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Os5Y?%t  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  SS);*;44>?IIST^u^uhhrWsW} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggWW G G  77''o o""$W%W'')?*?,,.'/'11346688:k;k==?O@OBBD3E3GGIJLLNNPmQmSSUQVQXXZ5[5]]_`bbddfogoiikSlSnnp7q7ssuvxxzz|q}qUUDŽdž77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS3S 3SSSG:SG:S~S~S[S[SSpS#S#SSSRS`RS=[S=[SS$ SS S S S/S /SS SAlS AlSeS eS5S5SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Os@dFjLp  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  \\)H*H44>2?2IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hh\\ P P  DD88,,   "#%%'(*v*v,,.b/b113N4N668:9:;;=&>&@@BCEEGGIrJrLLNZOZQQSBTBVVX*Y*[[]^``bbdnenggiVjVlln@o@qqs(t(vvxy{{}}llTTȇȉ<<$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSA˃SA˃SS SSS:S:SSSSS1Sp1SSSSSRS`RS=[S=[SS$ SS S S S;S ;SS SMlS MlSS SeSeS S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a- b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1H} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a- b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1H} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a- b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1H k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 [Y}W{Uy  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  UU)G*G44>7?7IIS#T#^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccUU G G  99++  "#%z%z'')l*l,,.^/^113P4P668B9B;;=4>4@@B&C&EEGHJJLMO}O}QQSkTkVVXYYY[[]G^G``b3c3eeghjjlmooqqsjtjvvxPyP{{}6~6rrVVȑȓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS-S -SSSA:SA:S~S~S[S[SSpSSSSSSS`SS\S\SS$ SS S S SS SS S)lS )lS S  S߉S߉SeS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Os /?cs  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  VV)>*>44>$?$IIST^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&& ~ ~""$j%j'')R*R,,.:/:113"4"668 9 ;~;~==?f@fBBDNENGGI4J4LLNOQQSSUnVnXXZR[R]]_6`6bbdeggiikplpnnpXqXssu>"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSA˃SA˃SS SSS:S:SSSSS1Sp1SSSSSQS`QSZSZSS$ SS S S S#S #SS S5lS 5lS#S #S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c/ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1H} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c/ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1H} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c/ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1H k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Os5Y?<  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  QQ)?*?44>+?+IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccQQ ? ?  --  n n""$\%\'')J*J,,.8/8113&4&6689;;=>@y@yBBDgEgGGISJSLLN=O=QQS'T'VVXY[[]]_p`pbbdVeVggi?IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%ss] ]""$G%G'')1*1,,./11346z6z88:d;d==?J@JBBD0E0GGIJLLNNPoQoSSUUVUXXZ;[;]]_!`!bbdeggiiklllnnpRqRssu8v8xxz{}}ww[[͉͋??## b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17L_SG˃SG˃SS SSS:S:SSSSS1Sp1SSSSSeQS`eQSZSZSS SM S| M SAS ASi~S i~SGNS GNSS SމSމSKS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j6 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Q} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j6 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Q} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j6 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Q k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1TxDh4X;_  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  OO)I*I44>7?7IIS#T#^^hiss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__OO ? ?  //w w""$q%q'')a*a,,.Q/Q113A4A668191;;=!>!@@BCEEGHJwJwLLNcOcQQSOTOVVX;Y;[[]'^'``bceeggiujullnmomqqsYtYvvxEyE{{}1~1 kk b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17L_SʃSʃS'S 'SSS;:S;:S~S~SISISSpSSSSSQS`QSZSZSS SM S| M SS S-~S -~S NS  NS S  S݉S݉S+S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1~ zvr  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  NN)4*444>?IIST^u^uhhrUsU} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%qqW W""$=%='')#*#,,. / 1|1|335b6b88:H;H==?*@*BBD E G}G}IIK_L_NNPAQASSU#V#XXZ[]v]v__adbdddfFgFiik(l(nnp q s{s{uuw]x]zz|?}?!!qqQQ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17L_SG˃SG˃SS SSS:S:S#S#SSS7Sp7SSSSSSS`SS\S\SS SM S| M S)S )SQ~S Q~S/NS /NS5S 5S+މS+މSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1R} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1R} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1R k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1OsZ~DhSw  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  KK)?*?44>)?)IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__KK 7 7  ##qq] ]""$Q%Q'')=*=,,.)/)11346689;w;w==?c@cBBDOEOGGI9J9LLN!O!QQS T V}V}XXZe[e]]_M`Mbbd5e5ggijllnoqqssumvmxxzU{U}}==%% ii b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations1YSʃSʃSS|SSSSS"S|"S%&SH%&S9S9Sc~SPc~S SX SSSSPSSSwSwSS$ SS, SS SS$ SS S]DS| ]DSRSXRSC[S|C[SSSDžSxDžSS\SSS'S'S/S8/SSSAlSAlSySySGSlGSSSS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1A bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1A  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1A bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`)~)Mt+vRvv n0TTx2f2V b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@8\\[\5Y3\| \Vz\1U /]x]] ]   ]   ]   ]   ]  ]]_c g kosw{!%) - 159= A  b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d\]8&op8L(`<v  T  2   n L *fD bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa QQ)G*G44>1?1IIS%T%^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaUU I I   ww!!   ""$y%y'')e*e,,.Q/Q113=4=668)9);;=>@@BBDqEqGGIYJYLLNAOAQQS)T)VVXY[[]]_m`mbbdaeaggiIjIlln1o1qqstvvxy{u{u}}]]тфEE,,# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYS_˃S_˃SSdSSlS!S!S%S%S:S@:S5S5SgSgSS@S}S }SISISSSS SwS wSSd S=CS =CSORS ORS[S [SӚSӚS˅S˅SgSgSSS(S (SSSSpS)lSt)lSySTyS S S߉S$߉S_S(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Q}" bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Q}" bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Q}"  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`%I^ /)Mb9-QhC3Wl)M b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@>l$s= Y}7[?cUy%I /s~&f D>b\   z  -  ' K  E i  c4.RNrp%#GEig><`^53WUy w !,!!!*"N"""L#p##$ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dliW8B-VnYD/    v c P =*q bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  NN)8*844> ? IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%ss] ]""$%''))+r,r..01346z6z88:d;d==?L@LBBD2E2GGIJLLNNPqQqSSUWVWXXZ=[=]]_#`#bbdeggiiknlnnnpTqTssu:v:xxz { }}yy__щыCC''  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1YSʃSʃSS|SSSSS"S|"S%&SH%&S9S9S]~SP]~S SX SSSSPSSSqSqSS$ SS, SS SS$ SS S]DS| ]DSSSXSS\S|\SSSхSxхSgS\gSSS*S*SS8SSS)lS)lSySyS Sl S߉S߉SGS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1A bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1A  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1A bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$*NkZ~ 0Ae}+;_aAe3i b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@.R)M{$HvC Rvq,P >  *  t  l N r  9 ( L  g  & o  4  I m b /]*X|%Sw Jn?c 4X)MwBl 7  a!!","""V#z##!$$$K%o%%&&&@'d'' (((5)Y))* b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d J#\%(-  2  5  8  ;  >ADGJMP bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa MM)=*=44>#?#IIST^^hhrisi} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaQQ A A ttdd  i i""$Y%Y'')A*A,,.)/)11346688:m;m==?S@SBBD7E7GGIJLLNNPqQqSSUUVUXXZ9[9]]_`bbd e ggiikclcnnpGqGssu+v+xxz{}}eeׄ׆II++ # b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYS_˃S_˃SSdSSlS!S!S%S%S:S@:S;S;SgSgSS@S}S }SOSOSSSS SwS wSSd S=CS =CSSS SS]S ]SYSYSхSхSSSSS*S *SSSSpSlStlSySTyS S SW߉S$W߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Q~" bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Q~" bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Q~"  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`)1Udl"FTx"F:?cwDh b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@0T.R,P*NLp~2V( L  <  " | q & J  W {  z = a $ H # G  -x"Fv DtBr@l8d 0\(Tx LpD h  !!!<"`""###4$X$$%%%,&P&&&|''$(H(((t))*@* b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d %t$C*0a6  <  @  D  H  LPTX\`d bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  NN)4*444>?IIST^w^whhrYsY} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%qqW W""$%'{'{))+a,a..0|1|335b6b88:H;H==?,@,BBDEGGIIKaLaNNPCQCSSU%V%XXZ[]x]x__afbfddfHgHiik*l*nnp q s}s}uuw_x_zz|A}A##uuUUŐŒ  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSʃSʃS-Sd-SSlS!S!S %S %S;:S@;:S~S~SmSmSS@SS SSSSSS S}S }SSd SCCS CCSQS QS[S [S[S[SŅSŅSmSmSSS&S &S5S5S Sp SGlStGlSySTySSS_S$_SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  Q" bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  Q" bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  Q"  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`**N}s(hMqqRv'KKo<`%%I b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@;SwAed/SBfA D /"mpKo ^+OLp:^( L   :   ( {   i   W {  Ei5Y'K= /!vhZ~Lp>b0T"F8*qc b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dSb&-yxLCrEk> h =   g  <  f; bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  SS)E*E44>7?7IzIzS#T#^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaSS E E  77))   ""$x%x'')j*j,,.\/\113N4N668@9@;;=2>2@@B$C$EEGHJJLMO}O}QQSkTkVVX%Y%[[]^``b3c3eeghjjlmooqqsjtjvvxPyP{{}6~6ttXXʑʓ  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSS˃SS˃SSdSSlS!S!S %S %S:S@:S/S/SmSmSS@SS SCSCSSSS S}S }SSd SCCS CCSaRS aRS[S [S嚅S嚅SqɅSqɅS-S-SSS(S (SSSSpS/lSt/lSySTyS)S)S߉S$߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}I b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Qc" bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}I b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Qc" bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}I  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Qc"  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$H\>$H\$H$H\.R$H\8\ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ 1b~ n`zBf$$H 0*w< Y}$H;_0T<`H l  T x  `  l   x  " .:$H4XDhTxdt  0@,P<`Lp\l| (  !8!!! b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d bZD'&tAjL.rT8    d  H  ,tX< bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  QQ);*;44>&?&IIST^P^Phhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$p p""$Z%Z'')D*D,,../.11346689;w;w==?a@aBBDKEKGGI6J6LLNOQQSTVzVzXXZb[b]]_J`Jbbd4e4ggijllnnpDqDssu,v,xxzH{H}}00xxbb  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSʃSʃS'Sd'SSlS!S!S %S %S;:S@;:S~S~SmSmSS@SS SSSSSS S}S }SSd SCCS CCSyRS yRS[S [SSSąSąSSSSS&S &SSSSpS/lSt/lSySTySSS߉S$߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  Q" bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  Q" bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  Q"  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$/Sh:99]*NCn>b)TxRv b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Qu;_%I3nX|Bf,P -x: Vz $ 4 X u    6 _   I m _  3 W = a   A  ?  +|fPt:^$H2mW{Ae+O9#t ^Hl2V@ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dQ"sDfY7$oY:*{L   n  ?   a 2T bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  OO)=*=44>+?+IkIkST^^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaOO = =  ++~~l l""$Z%Z'')H*H,,.6/6113$4$6689;;=>@w@wBBDeEeGGISJSLLN=O=QQS'T'VgVgXXZQ[Q]]_;`;bbdVeVggi!?!IIS T ^G^Ghhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&rr\ \""$F%F'')0*0,,./11346y6y88:c;c==?M@MBBD7E7GGI!J!LLNOQzQzSSU`V`XXZF[F]]_,`,bbdeggiikjljnnpqssuuwoxoz{}}llބކRRʼnŋ88  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃS-SL-SSTS!S!S%S%SA:S(A:S~S~SsSsS S( SSSSSSSS SS SSL SICS ICSSS SSQ\S Q\SSSSSSS5SD5S S SGlS GlSSSkS0kSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Q" bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Q" bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Q"  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`++Or6Z%HllBrBf4<Lp b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@4cEi'K -z\>b bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1]*wD } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  QQ)A*A44>1?1IIST^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaQQ A A  11!! y y""$i%i'')Y*Y,,.I/I113949668)9);;=>@@B C EEGGIqJqLLN]O]QQSITIVVX5Y5[[]!^!``b c eeggiojolln[o[qqsGtGvvx3y3{{}~ mmYYϑϓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SM˃SM˃SSLSSTS!S!S%S%S:S(:S)S)SsSsS S( SSS=S=SSSS SS SSL SICS ICSSS SS ]S  ]SSSSSSSSSSS)SD)SSS;lS ;lSSSSS#S0#SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1r> b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 QG& bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1r> b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 QG& bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1r>  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 QG&  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`#GZ~ChVz@?Rv<Os8\ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$oMq+O -x Vz4X6 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1Ka, } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  RR)@*@44>0?0IIST^^histst} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddRR @ @  ..  o o""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGIXJXLLNFOFQQS4T4VVX"Y"[[]^``bbdueuggicjcllnLoLqqs0t0vvxy{{}}jj܂܄NN22 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃS-SL-SSTS!S!S%S%SA:S(A:S~S~SsSsS S( SSSSSSSS SS SSL SICS ICSQS QS[S [S[S[SSSSSSDSSS/lS /lS#S#S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Q" bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Q" bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Q"  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$u4Xe_BBfIm,,P}3W:g b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ D&s U y 7 [  =  l  N r  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1^+xE } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  MM)9*944>%?%IIS T ^^hhrisi} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%ss_ _""$K%K'')7*7,,.#/#11346688:q;q==?]@]BBDIEIGGI5J5LLNOQQSTVyVyXXZa[a]]_I`Ibbd1e1ggijllnoququssu]v]xxzE{E}}--qqYY b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SS˃SS˃SSLSSTS!S!S%S%S:S(:S/S/SsSsS S( SSSCSCSSSS SS SSL SICS ICSaRS aRS[S [S嚅S嚅SCSCSSSSDSSS)lS )lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vB b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 QH' bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vB b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 QH' bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vB  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 QH'  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`'5Y^7#&JdB.RUy(L2V b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@_=a ?  h F j $ H  & q  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1wB X# } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  LL)6*644>"?"IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbLL 6 6      iiS S""$=%='')'*',,./11335p6p88:Z;Z==?D@DBBD.E.GGIJLLNOQyQySSUcVcXXZM[M]]_7`7bbd!e!ggi j llnnpdqdssuHvHxxz,{,}}ff؉؋JJ.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS3SL3SSTS!S!S%S%SG:S(G:S~S~SySySS(SSSSS#S#SS SS SSL SOCS OCSRS RS\S \SWSWSSSSS)SD)SSS;lS ;lSeSeS5S05SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 S$ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 S$ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 S$  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`,,Pq'K"CggOs99]~  /w / b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@+z`Fj,P6kQu bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1V%tC } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  OO)=*=44>+?+IIST^^hhrtst} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaOO = =  ++~~l l""$Z%Z'')H*H,,.6/6113$4$6689;;=>@w@wBBDeEeGGISJSLLN=O=QQS'T'VVXY[[]]_p`pbbdZeZggiDjDlln.o.qqstvvxy{w{w}}aaւքKK55 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSG˃SG˃SSLS ST S!S!S#%S#%S:S(:S#S#SSSS(SSS7S7SSSS SS SSL S[CS [CSSS SS\S \S S SSS1S1S/SD/SSSAlS AlSeSeS5S05SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g3 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 QX1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g3 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 QX1 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g3  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 QX1  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`"FX|>X|Pt&6Hl .Vz@d$Tx b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$zn bVzJn>b2V&J bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1V,X } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  \\)H*H44>4?4IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hh\\ P P  DD88,,   "#%%'(*v*v,,.b/b113N4N668:9:;;=&>&@@BCEEGGItJtLLN\O\QQSDTDVVX,Y,[[]^``bbdpepggiXjXlln@o@qqs(t(vvxy{{}}llTTȇȉ<<$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS3SL3SSTS!S!S%S%SG:S(G:S~S~SsSsS S( SSSSS#S#SS SS SSL SICS ICSQS QS[S [SaSaS͵S͵S'S'SSDSSS)lS )lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 R# bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 R# bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 R#  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`)>b|lf,PUyPt>Cg:^,1U b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@7[A ' v \ B f ( L  4  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1a0N } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  NN)8*844>"?"IIST^{^{hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&tt` `""$K%K'')5*5,,./113 4 6~6~88:h;h==?R@RBBD 2  & |  p  d  X | Lp bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1.Z0 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  VV)>*>44>&?&IIS T ^|^|hhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&& ~ ~""$j%j'')R*R,,.:/:113"4"668 9 ;~;~==?f@fBBDNENGGI6J6LLNOQQSSUpVpXXZT[T]]_8`8bbdeggijlrlrnnpVqVssu:v:xxz{}}ttXXʉʋ<<  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS9SL9SSTS!S!S5%S5%SM:S(M:S~S~SSS/S(/SSSũSũS)S)SS SS S%SL %SmCS mCSIRS IRS[S [S͚S͚SѮSѮS+S+S)SD)SSS;lS ;lSGSGSS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  i@ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  i@ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  i@  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Qu --Qpe>bb 0y0Ts'"; b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$weSwAe/SA / bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1S&yL } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  UU)A*A44>+?+IIST^^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeYY M M  AA55%%  ""$s%s'')_*_,,.K/K113747668#9#;;=>@@BBDkEkGGISJSLLN;O;QQS#T#VVX Y [[]]_g`gbbdOeOggi7j7llnoqqstv{v{xxzc{c}}KK33ss b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSA˃SA˃SSLSSTS!S!S%S%S:S(:SSSySySS(SSS1S1SSSS SS SSL SOCS OCSQS QS[S [SaSaSSSaSaSSDSSS/lS /lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~\( b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 QE% bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~\( b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 QE% bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~\(  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 QE%  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`!EQuVz;_:Ko%I .?c3"3Wx b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$ .8B(L2V<`Fj bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1aB# } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  PP):*:44>"?"IIST^{^{hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))uu_ _""$I%I'')3*3,,./11346|6|88:f;f==?N@NBBD4E4GGIJLLNOQsQsSSUYVYXXZ?[?]]_%`%bbd e g~g~iikdldnnpJqJssu0v0xxz{}}ooUUljNj99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS9SL9SSTS!S!S5%S5%SG:S(G:S~S~SSS/S(/SSSSS#S#SS SS S%SL %SmCS mCSQS QS7[S 7[SSSESESKSKSSDSSS)lS )lS S S߉S0߉SGS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  h> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  h> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  h>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Sw .Hlv=a:^h:^,PZ~DhBLp b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@p ^L p : ^ ( L  :  ( { i bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1rEk } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  QQ)9*944>?IIST^u^uhhrYsY} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%%m m""$U%U'')=*=,,.%/%113 4 6688:i;i==?O@OBBD3E3GGIJLLNNPmQmSSUQVQXXZ5[5]]_`bbddfogoiikSlSnnp7q7ssuvxxzz|q}qUUDŽdž99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSG˃SG˃SSLSSTS!S!S%S%S:S(:SSSSSS(SSS1S1SSSS SS S SL  SUCS UCSRS RS=[S =[SSSSSSS SD SSSlS lS S S]߉S0]߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`, b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 QE& bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`, b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 QE& bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`,  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 QE&  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`(Z~<`d2Vz6ZwRv,ImqNr? b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@PtZ ~  d n  x ! +5? bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1eF'i } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  NN)4*444>?IISS]k^khhrMsM} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))qqW W""$=%='')#*#,,. / 1|1|335b6b88:H;H==?,@,BBDEGGIIKaLaNNPCQCSSU%V%XXZ[]x]x__aZbZddf;?;IIS'T'^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeWW I I  ;;--  "#%|%|'')n*n,,.`/`113R4R668D9D;;=6>6@@B(C(EEGHJJL M OOQQSoToVVX]Y][[]K^K``b7c7eeg!h!jjl m o~o~qqsdtdvvxJyJ{{}0~0nnRRđē b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS;˃S;˃SSLSSTS!S!S%S%S:S(:SSSSSS(SSS%S%SSSS SS S SL  SUCS UCSRS RS[[S [[SSSKSKSSS/SD/SSSAlS AlS;S;S S0 SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1sQ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 QQ# bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1sQ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 QQ# bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1sQ  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 QQ#  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Qu@Lp*X|6Z{^ De .bOs b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$ D@d`40TPtp bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1lXD0 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  UU)?*?44>*?*IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhXX H H  88((t t""$^%^'')H*H,,.2/211346689;{;{==?e@eBBDOEOGGI:J:LLN"O"QQS T V~V~XXZf[f]]_N`Nbbd6e6ggijllnoqzqzssubvbxxzJ{J}}22zzdd b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS?SL?SSTS!S!S%S%SM:S(M:S~S~SSSS(SSSũSũS)S)SS SS S SL  SUCS UCSRS RS[S [S'S'SSSSS)SD)SSS;lS ;lS)S)S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 R# bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 R# bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 R#  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Qu3Rv $Hl}{9>bqAe4Xgm b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ D$ H ( L , P 0 T 4 X 8\<` bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1V4pN } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>/?/IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIWJWLLNAOAQQS+T+VVXY[[]]_t`tbbdZeZggi@j@lln$o$qqstvwvwxxzY{Y}};;qqSS b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS;˃S;˃SSLSSTS!S!S%S%S:S(:SSSSSS(SSS+S+SSSS SS S SL  SUCS UCSQS QSZS ZSCSCSKSKS/S/S#SD#SSS5lS 5lS S S߉S0߉SkS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uU! b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 QQ# bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uU! b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 QQ# bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uU!  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 QQ#  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Qu+Bfm;_=Dh%I?[31U b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ $ D @ d `  40TPtp bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1t`L } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  QQ);*;44>%?%IIS T ^|^|hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **vv` `""$J%J'')4*4,,./11346}6}88:g;g==?Q@QBBD;E;GGI%J%LLN O Q~Q~SSUdVdXXZJ[J]]_0`0bbdeggiikjljnnpNqNssu2v2xxz{}}llބކRRʼnŋ88 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃSESLESSTS!S!S%S%SS:S(S:S~S~SSSS(SSS˩S˩S/S/SS SS S SL  SUCS UCSRS RS\S \ScScS'S'SSSGSDGSSSYlS YlSSS}S0}S%S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 R# bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 R# bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 R#  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Qu /1Un4XZ~+O D] g) b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$@6ZPtj6,PFj bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1kT=& } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  UU)E*E44>5?5IIS!T!^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%% } }""$m%m'')]*],,.M/M113=4=668-9-;;=>@@B C EEGGIuJuLLNaOaQQSMTMVVX9Y9[[]%^%``bceeggisjslln_o_qqsKtKvvx7y7{{}#~#qq]]ӑӓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqS5˃S5˃SSLSSTS!S!S%S%S:S(:S S SSSS(SSSSSSSS SS S SL  SUCS UCSQS QS[S [SaSaSSSSS;SD;SSSMlS MlSeSeS5S05SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1hF b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 QG" bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1hF b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 QG" bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1hF  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 QG"  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Qu33WAe=h2V#Nrt&J 4XX| b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$6Zl+=as2Dhz 9  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1wne\ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  VV)D*D44>4?4IIS"T"^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhVV D D  22  s s""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGI\J\LLNJOJQQS8T8VVX&Y&[[]^``bceyeyggigjgllnPoPqqs4t4vvxy{{}}nnRRćĉ66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃS?SL?SSTS!S!S%S%SS:S(S:S~S~SSSS(SSS˩S˩S/S/SS SS S SL  SUCS UCSRS RS[S [S9S9SSSSS/SD/SSSAlS AlS5S5SS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 R# bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 R# bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 R#  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Qu6X|  /Rv=)Lp#ImFj|y b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@`  z  + ! E ; _ Uyo : bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1xaJ3 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  QQ)=*=44>)?)IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))wwc c""$O%O'');*;,,.'/'11346688:u;u==?a@aBBDMEMGGI9J9LLN!O!QQS T V}V}XXZe[e]]_M`Mbbd5e5ggijllnoqyqyssuavaxxzI{I}}11uu]] b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqS5˃S5˃SSLSSTS!S!S%S%S:S(:SSSSSS(SSS%S%SSSS SS S SL  SUCS UCSQS QS[S [SISISSSSS)SD)SSS;lS ;lS S S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1jJ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 QF" bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1jJ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 QF" bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1jJ  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 QF"  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Qu'9]`=a79])M7Z~95Y b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ K o  @ R v #GY}*N bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1SJA8/  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  PP):*:44>&?&IIST^^hhrfsf} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffPP : :  $$mmW W""$A%A'')+*+,,./11335t6t88:^;^==?H@HBBD2E2GGIJLLNOQ}Q}SSUgVgXXZQ[Q]]_;`;bbd%e%ggijllnnphqhssuLvLxxz0{0}}jj܉܋NN22 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃSKSLKS ST S!S!S#%S#%SY:S(Y:S~S~SSSS(SSSѩSѩS5S5SS SS SSL S[CS [CSRS RS[S [S!S!SϰSϰSSSASDASSSSlS SlSwSwSGS0GSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 R# bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 R# bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 R#  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Qu&&Jc %IyRv 1 >Hl~5Y] b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$0T`(LX| DPt  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1th\P } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>/?/IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIWJWLLNAOAQQS+T+VVXY[[]]_t`tbbd^e^ggiHjHlln2o2qqstvvxy{{{{}}eeڂڄOOćĉ99## b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqS)˃S)˃SSLS ST S!S!S#%S#%S:S(:SSSSSS(SSSSS}S}SS SS SSL S[CS [CSQS QS[S [SISISSSSSASDASSSSlS SlS}S}SMS0MSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]; b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 QQ# bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]; b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 QQ# bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1];  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 QQ#  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Qu&&Js;_,Uyy%I7[[3==a b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$Lp>b0T|"Fn 8  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  ``)L*L44>8?8IIS T ^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ll`` T T  HH<<00$$  "#%%'(*z*z,,.f/f113R4R668>9>;;=*>*@@BCEEGHJxJxLLN`O`QQSHTHVVX0Y0[[]^``bcetetggi\j\llnDoDqqs,t,vvxy{{}}ppXẊ̉@@(( b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃSESLES ST S!S!S#%S#%SY:S(Y:S~S~SSSS(SSSѩSѩS5S5SS SS SSL S[CS [CSRS RS[S [S-S-SMSMS}S}S/SD/SSSAlS AlSSS߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 R# bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 R# bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 R#  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Qu /Jnz }Hlw7Fju?cDhsk b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@  < H l  x  4 @dp,8\h bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1D8,   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  RR)<*<44>&?&IIS T ^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **xxd d""$O%O'')9*9,,.#/#113 4 6688:l;l==?V@VBBD@E@GGI*J*LLNOQQSSUiViXXZO[O]]_5`5bbdeggijltltnnpZqZssu@v@xxz&{&}} ee؉؋KK11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqS/˃S/˃SSLS ST S!S!S#%S#%S:S(:S S SSSS(SSSSSSSS SS SSL S[CS [CSQS QS[S [S[S[SSSSS)SD)SSS;lS ;lSSS߉S0߉S}S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_?  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 QQ# bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_?  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 QQ# bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_?   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 QQ#  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Qu&/ShAe 14X|1U3Z~~!E8\ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@`  * R v  Dh6Z(Lt> bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1    } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  ZZ)B*B44>*?*IIST^^hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjZZ J J  ::**   ""$n%n'')V*V,,.>/>113&4&6689;;==?j@jBBDRERGGI:J:LLNOQQSTVtVtXXZX[X]]_<`v>xxz"{"}}xx\\Ή΋@@$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}SʃSʃScScSS$Sw:Sw:S~S~SgS$gS梅S梅SSSSSS$SQSQS7[S47[SçST çSk S k SMS MS%S %S_lS _lSkS kS;S;SS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1 Ai b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1 Ai b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1 Ai b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 kUy?c  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10zz  jj)V*V44>>?>IIS&T&^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1zznn b b  VVJJ::&&  "#%%'')t*t,,.`/`113L4L668898;;=">"@@B C E~E~GGIfJfLLNNONQQS6T6VVXY[[]^`z`zbbdbebggiJjJlln2o2qqstvvxy{v{v}}^^҂҄FF++ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S)˃S)˃SS SSS:S:S~S~SCSCSSpSwSwSSS'SS`'SSc\Sc\SçS$ çSk S k S;S ;SS SMlS MlS S  S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q/ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1]} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q/ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1]} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q/ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1] k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 @dt2Bf  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10rr  ]])G*G44>-?-IIST^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1rr^^ J J  66""  l l""$V%V'')@*@,,.*/*11346688:s;s==?Y@YBBD?E?GGI%J%LLN O Q~Q~SSUdVdXXZJ[J]]_0`0bbdeggiikolonnpUqUssu;v;xxz!{!}}zz^^ЉЋBB&& b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}SʃSʃS]S]SS$Sq:Sq:S~S~SgS$gS驄S驄SMSMSS$SCRSCRS[S4[SçST çSk S k S5S 5S S  SGlS GlS S  S߉S߉SkS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bg b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bg b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Bg b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 *&"|  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10xx  dd)L*L44>0?0IIST^^hhrhsh} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1xxhh X X  HH88$$   ""$h%h'')P*P,,.8/8113 4 6689;|;|==?`@`BBDDEDGGI(J(LLN O Q~Q~SSUbVbXXZF[F]]_*`*bbdeggiikdldnnpHqHssu,v,xxz{}}ff؄؆HH**  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S)˃S)˃SS SSS:S:SSSCSCSSpS}S}SSS3SS`3SSo\So\SçS$ çSk S k S)S )SS S;lS ;lS S  S{߉S{߉S#S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q1 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1\} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q1 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1\} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q1 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1\ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Cg{;Os  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  ]])C*C44>%?%IIST^x^xhhrXsX} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1tt`` L L  88$$  f f""$L%L'')2*2,,./11335q6q88:W;W==?9@9BBDEGGIIKnLnNNPPQPSSU2V2XXZ[]]__agbgddfIgIiik+l+nnp q s~s~uuw`x`zz|B}B$$ttTTĐĒ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃS]S ]SSSk:Sk:S~S~SaSaS㩄Sp㩄SGSGSSSQS`QS+[S+[SS$ Se S e SYS YS1S 1SklS klSS SSS+S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 zdNr8\  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10vv  hh)Z*Z44>J?JIIS6T6^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvhh Z Z  LL>>00""  "#%%'(**,,.q/q113c4c668U9U;;=G>G@@B9C9EEG+H+JJLMOOQRT~T~VVXlYl[[]Z^Z``bFcFeeg0h0jjlmooqrtstsvvxYyY{{}?~?%% {{__ёѓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS/˃S/˃SS SSS:S:SSS1S1SSpS}S}SSSRS`RS\S\SS$ Se S e SAS ASS SSlS SlSMS MSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z8 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Q} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z8 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Q} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z8 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Q k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 1UQuq%  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  ^^)H*H44>1?1IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qqaa Q Q  AA11 } }""$g%g'')Q*Q,,.;/;113%4%6689;;==?n@nBBDXEXGGIAJALLN)O)QQSTVVXXZm[m]]_U`Ubbd=e=ggi%j%lln o qqssuivixxzQ{Q}}99!! ii b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃSWS WSSSk:Sk:S~S~SaSaS㩄Sp㩄SGSGSSSORS`ORS[S[SS$ Se S e SAS ASS SSlS SlSAS ASSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  - - - -  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  bb)P*P44>>,,  ""$m%m'')[*[,,.I/I113747668%9%;;=>@@BCExExGGIdJdLLNNONQQS8T8VVX"Y"[[] ^ ``bbdgegggiMjMlln1o1qqstvvxxzf{f}}HH** ||^^ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS/˃S/˃SS SSS:S:S S S1S1SSpSSSSSSS`SS\S\SS$ Se S e S5S 5S S  SGlS GlSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z: b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Q} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z: b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Q} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z: b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Q k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 X|&Vz$Tx  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  \\)F*F44>.?.IIST^^hhrisi} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qq]] I I  55!!  k k""$U%U'')?*?,,.)/)11346688:r;r==?\@\BBDFEFGGI.J.LLNOQQSSUmVmXXZS[S]]_9`9bbdeggijlslsnnpWqWssu;v;xxz{}}uuYỶ̋??%% b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃSWS WSSSe:Se:S~S~SgSgSݩSpݩSASASSSQS`QS%[S%[SS$ Se S e SSS SS+S +SelS elSS SSS7S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 u_Im3W  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10rr  bb)R*R44>@?@IIS,T,^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1rrbb R R  BB22""  "#%z%z'')j*j,,.Z/Z113J4J668:9:;;=*>*@@BCEEG H JJLLNlOlQQSXTXVVXDYD[[]0^0``bceeghj~j~llnjojqqsVtVvvxByB{{}.~.||hhޑޓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS/˃S/˃SS SSS:S:S S S%S%SSpSSSSSRS`RS[S[SS$ Se S e SAS ASS SSlS SlSkS kS;S;SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1cA  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1F} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1cA  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1F} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1cA  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1F k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 C)M3W=a  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  [[)I*I44>7?7IIS%T%^^h i s{s{} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mm[[ I I  77%% x x""$f%f'')T*T,,.B/B1130406689;;= > @@BBDqEqGGI_J_LLNMOMQQS;T;VVX)Y)[[]^``bce|e|ggijjjllnSoSqqs7t7vvxy{{}}qqUULJlj99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃSQS QSSSe:Se:S~S~SgSgSݩSpݩSASASSS=RS`=RSy[Sy[SS$ Se S e S;S ;SS SMlS MlSGS GSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 *&"|  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  \\)H*H44>2?2IIST^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp\\ H H  44    n n""$Z%Z'')F*F,,.2/21134668 9 ;;==?l@lBBDXEXGGIBJBLLN*O*QQSTVVXXZn[n]]_V`Vbbd>e>ggi&j&llnoqqssujvjxxzR{R}}::"" ~~ff b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS5˃S5˃SS SSS:S:S S S%S%SSpSSSSSSS`SS\S\SS$ Se S e S5S 5S S  SGlS GlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1cC b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1E} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1cC b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1E} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1cC b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1E k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 [+_ /c  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  WW)A*A44>+?+IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mmWW A A  ++tt^ ^""$H%H'')2*2,,./11346{6{88:e;e==?O@OBBD9E9GGI#J#LLN O QQSSUlVlXXZV[V]]_@`@bbd*e*ggijllnnpmqmssuQvQxxz5{5}}ooSSŎŐ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃSKS KSSS_:S_:S~S~S[S[SשSpשS;S;SSSQS`QS%[S%[SS$ S_ S _ SAS ASS SSlS SlSwS wSGSGSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1` b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1` b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1` b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 waKo5Y  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  \\)J*J44>6?6IIS T ^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nn\\ J J  88&& y y""$g%g'')U*U,,.C/C1131416689;;= > @@BBDrErGGI^J^LLNHOHQQS2T2VVXY[[]^`{`{bbdeeeggiOjOlln9o9qqs#t#vvx y {{}}llVVˇˉ@@** b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS5˃S5˃SS SSS:S:SSS1S1S%Sp%SSSSSyRS`yRS[S[SS$ S_ S _ SAS ASS SSlS SlSwS wSGSGSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1lJ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Z} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1lJ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Z} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1lJ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Z k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 >&J2V:^  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  aa)M*M44>7?7IIST^^his{s{} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mmaa U U  II==11%%  "#%%'(*{*{,,.g/g113S4S668?9?;;=+>+@@BCEEGHJwJwLLN_O_QQSGTGVVX/Y/[[]^``bbdsesggi[j[llnCoCqqs+t+vvxy{{}}ooWWˇˉ??'' b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃSKS KSSSY:SY:S~S~S[S[SѩSpѩS5S5SSSCRS`CRS[S[SS$ S_ S _ S/S /SS SAlS AlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  - - - -  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  YY)C*C44>+?+IIST^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mmYY E E  11  k k""$V%V'')@*@,,.*/*11346688:s;s==?]@]BBDGEGGGI/J/LLNOQQSSUnVnXXZT[T]]_:`:bbd e ggijlylynnp_q_ssuEvExxz+{+}}jj݉݋PPÎÐ66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS;˃S;˃SS SSS:S:SSS7S7S%Sp%SSSSSSS`SS\S\SS$ S_ S _ S)S )SS S;lS ;lS S  S߉S߉SwS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1lL b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1[} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1lL b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1[} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1lL b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1[ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 X|&Vz$Tx  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  ]])E*E44>+?+IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mm]] M M  ==--   ""$q%q'')Y*Y,,.A/A113)4)6689;;==?m@mBBDUEUGGI;J;LLNOQQSTVuVuXXZY[Y]]_=`=bbd!e!ggijlwlwnnp[q[ssu?v?xxz#{#}}yy]]ωϋAA%% b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSES ESSSY:SY:S~S~SgSgSѩSpѩS5S5SSSRS`RS=[S=[SS$ SY S Y S5S 5S S  SGlS GlSMS MSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Qu;_<`Fj  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  ^^)J*J44>2?2IIST^^histst} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nnbb V V  JJ>>..  "#%|%|'')h*h,,.T/T113@4@668,9,;;=>@@BBDrErGGIZJZLLNBOBQQS*T*VVXY[[]]_n`nbbdVeVggi>j>lln&o&qqstvvxxzj{j}}RRƂƄ::xx b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS;˃S;˃SS SSS:S:SSSSS+Sp+SSSSS+RS`+RSg[Sg[SS$ SY S Y SS SS S/lS /lS S  S߉S߉SkS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uS b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1F} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uS b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1F} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uS b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1F k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 `d hn  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  QQ);*;44>!?!IIST^z^zhhr^s^} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **vv` `""$J%J'')4*4,,./11346}6}88:g;g==?M@MBBD3E3GGIJLLNNPrQrSSUXVXXXZ>[>]]_$`$bbd e g}g}iikclcnnpIqIssu/v/xxz{}}nnRRĉċ66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSES ESSSS:SS:S~S~SgSgS˩Sp˩S/S/SSS%RS`%RSa[Sa[SS$ SY S Y SS SS S/lS /lS S  S߉S߉SMS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 QuTx^h  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  XX)@*@44>$?$IIST^z^zhhr\s\} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ll\\ L L  <<,, t t""$\%\'')D*D,,.,/,11346688:p;p==?T@TBBD8E8GGIJLLNOQrQrSSUVVVXXZ:[:]]_`bbdegtgtiikXlXnnp?IISS]l^lhhrLsL} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,ttZ Z""$@%@'')&*&,,. / 11335e6e88:K;K==?-@-BBDEGGIIKbLbNNPDQDSSU&V&XXZ[]y]y__a[b[ddf=g=iiklnnpqsrsruuwTxTzz|6}6hh؋؍HH b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS?S ?SSSS:SS:S~S~SOSOS˩Sp˩S/S/SSSRS`RSC[SC[SS$ SY S Y S;S ;SS SMlS MlSS SeSeS S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Qu<`FjPt  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  \\)N*N44>>?>IIS*T*^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jj\\ N N  @@22$$  "#%%'')s*s,,.e/e113W4W668I9I;;=;>;@@B-C-EEGHJJL M OOQQSrTrVVX`Y`[[]N^N``b:c:eeg$h$jjlmooqqsgtgvvxMyM{{}3~3ooSSőœ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSA˃SA˃SS SSS:S:SSS%S%S1Sp1SSSSSQS`QS%[S%[SS$ SY S Y S#S #SS S5lS 5lS/S /S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~\( b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1V} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~\( b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1V} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~\( b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1V k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 W{MqCg9]  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  RR)<*<44>%?%IIS T ^^hhrisi} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%%q q""$[%['')E*E,,.///11346689;x;x==?b@bBBDLELGGI5J5LLNOQQSTVyVyXXZa[a]]_I`Ibbd1e1ggijllnoququssu]v]xxzE{E}}--ss]] b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS9S 9SSSM:SM:S~S~SOSOSũSpũS)S)SSSRS`RS[S[SS$ SY S Y S#S #SS S5lS 5lS#S #S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Qu =Imy  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  VV)D*D44>0?0IIST^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhVV D D  22  s s""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIXJXLLNBOBQQS,T,VVXY[[]^`u`ubbd[e[ggiAjAlln%o%qqstvxvxxxzZ{Z}}<<ppRR b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSA˃SA˃SS SSS:S:SSS+S+S1Sp1SSSSSQS`QSZSZSS$ SY S Y SS SS S)lS )lS S  S߉S߉SeS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~^* b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1W} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~^* b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1W} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~^* b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1W k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Qu;_%I5  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  PP):*:44>"?"IIST^y^yhhr]s]} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))uu_ _""$I%I'')3*3,,./11346|6|88:f;f==?P@PBBD:E:GGI"J"LLNOQ{Q{SSUaVaXXZG[G]]_-`-bbdeggiikglgnnpKqKssu/v/xxz{}}iiۄۆMM33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS9S 9SSSM:SM:S~S~S[S[SũSpũS)S)SSSRS`RSO[SO[SS$ SS S S S;S ;SS SMlS MlSS SqSqSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 QuJnTx^  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  VV)F*F44>4?4IIS T ^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&& ~ ~""$n%n'')^*^,,.N/N113>4>668.9.;;=>@@BCEEGGItJtLLN`O`QQSLTLVVX8Y8[[]$^$``bceeggirjrlln`o`qqsLtLvvx8y8{{}$~$rr^^ԑԓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSG˃SG˃SS SSS:S:SSSSS1Sp1SSSSSQS`QSZSZSS$ SS S S S/S /SS SAlS AlSYS YS)S)SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e1 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1E} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e1 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1E} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e1 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1E k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Rv>b*N:  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  OO)=*=44>+?+IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaOO = =  ++~~l l""$Z%Z'')H*H,,.6/6113$4$6689;;=>@w@wBBDeEeGGISJSLLNAOAQQS/T/VVXY[[] ^ ``bbdpepggi^j^llnQoQqqs5t5vvxy{{}}ooSSŇʼn77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS3S 3SSSG:SG:S~S~S[S[SSpS#S#SSSESS`ESS\S\SS$ SS S S S#S #SS S5lS 5lS/S /S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Qu#[ -f8  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP)<*<44>&?&IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((vvb b""$N%N''):*:,,.&/&11346688:t;t==?`@`BBDLELGGI6J6LLNOQQSTVzVzXXZb[b]]_J`Jbbd2e2ggijllnoqzqzssubvbxxzJ{J}}22vv^^ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSG˃SG˃SS SSS:S:S#S#SSS7Sp7SSSSSQS`QS%[S%[SS$ SS S S SS SS S/lS /lS S  S߉S߉SwS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g3 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1E} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g3 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1E} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g3 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1E k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 QuImBf:^  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  KK)5*544>?IIS T ^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaKK 5 5    ~~hhR R""$<%<'')&*&,,./11335o6o88:Y;Y==?C@CBBD-E-GGIJLLNOQvQvSSU`V`XXZJ[J]]_4`4bbdeggijl}l}nnpiqissuMvMxxz1{1}}kk݉݋OO33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃS3S 3SSSA:SA:S~S~SISISSpSSSSSQS`QSZSZSS SM S| M S5S 5S]~S ]~S;NS ;NSkS kSaމSaމS S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1Qu@d,Pl  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  PP)@*@44>,?,IIST^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGITJTLLN>O>QQS(T(VVXY[[]]_q`qbbd[e[ggiEjElln;o;qqs%t%vvxy{{}}nnXX͇͉BB,, b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17L_SM˃SM˃SS SSS:S:S#S#SSS7Sp7SSSSSQS`QSZSZSS SM S| M S;S ;Sc~S c~SANS ANSqS qSgމSgމSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n: b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1T} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n: b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1T} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n: b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1T k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1Qu;_Jn#  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  UU)K*K44>5?5IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaUU I I  ==11%%  " # %%'(*y*y,,.e/e113Q4Q668=9=;;=)>)@@BCEEGHJuJuLLN]O]QQSETEVVX-Y-[[]^``bbdqeqggiYjYllnMoMqqs5t5vvxy{{}~yyaaՇՉII11 b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17L_SʃSʃS-S -SSSA:SA:S~S~SISISSpSSSSSRS`RS!\S!\SS SM S| M S#S #SK~S K~S)NS )NSS S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1Qu=Cg  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  MM);*;44>#?#IIS T ^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%ss_ _""$N%N'')8*8,,."/"113 4 6688:k;k==?U@UBBD?E?GGI'J'LLN O QQSSUfVfXXZL[L]]_2`2bbdeggiikqlqnnpcqcssuIvIxxz/{/}}nnTTǎǐ:: b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17L_SM˃SM˃SS SSS:S:S)S)SSS=Sp=SSSSSSS`SS]S]SS SM S| M S#S #SK~S K~S)NS )NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p< b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1U} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p< b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1U} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p< b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1U k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1Qu)M?c  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  QQ)A*A44>'?'IIS T ^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaQQ A A  11!! y y""$m%m'')U*U,,.=/=113%4%668 9 ;;==?i@iBBDQEQGGI7J7LLNOQQSSUqVqXXZU[U]]_9`9bbdeggijlslsnnpcqcssuGvGxxz+{+}}ee׉׋II-- b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations1YSʃSʃSS|SSSSS"S|"S%&SH%&S9S9Sc~SPc~S SX SSSSPSSSwSwSS$ SS, SS SS$ SS S]DS| ]DSaRSXaRS[S|[S嚅S嚅SʅSxʅSeS\eSqSqSo(So(SS8SSS/lS/lSySyS Sl S߉S߉SeS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1C bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1C  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1C bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`8Jn4:LpB<Nr,P>Pt:^ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@g},B4XJn`v% ;  - Q  C g  Y }   o   !E4h&JBf<`@Vzcp=a!=;1UKoe 0&J@dZ~ t% ?  5!Y!!!O"s""# b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dgN5jQ8l 5 o ~ X G A  *   |eN7  bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10]] RR)<*<44>$?$IIST^^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1]]II 5 5 u uaaMMwwa a""$K%K'')5*5,,./113 4 6~6~88:h;h==?P@PBBD6E6GGIJLLNOQuQuSSU[V[XXZA[A]]_'`'bbdeggiikrlrnnpXqXssu>v>xxz${$}} }}ccՉՋGG++# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYS_˃S_˃SSdSSlS!S!S %S %S:S@:S;S;SmSmSS@SS SOSOSSSS S}S }SSd SCCS CCSIRS IRS[S [S͚S͚SȅSȅSSSSS'S 'S)S)SSpS;lSt;lSySTySGSGSS$SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Sy& bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Sy& bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Sy&  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`,/Sk,1UX|6C[@ &J b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@`hpx ( 08@$ H  , P  4 X  < `  D h  LpTx\ /h5t;*A6#GB)M*N/S6Z5YBfNrZ~fr~( 4  !@! b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d`@ `@ `@^ " =    f {  H Z  *  nP2 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  UU)C*C44>-?-IIS!T!^^h i s}s}} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeYY M M  AA55%%  ""$?%?'')+*+,,./113949668%9%;;=>@@BBDmEmGGIUJULLN=O=QQS%T%VVX Y [[]]_i`ibbd]e]ggiEjElln-o-qqstvvxxzq{q}}YY͂̈́AA((  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1YSʃSʃSS|SSSSS"S|"S%&SH%&S9S9Sc~SPc~S SX SSSSPSSSwSwSS$ SS, SS SS$ SS S]DS| ]DSSSXSS#]S|#]SkSkSiԅSxiԅS% S\% S} S} S{+S{+SS8SSSlSlSySyS Sl S]߉S]߉SS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1C bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1C  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1C bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`&&Jmq Dhh)M>>b8Qu - b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Dh`8|0TLp $  h  @   8 \  Tx,p$H@d`\:^28r[$H5Y3|:^ Pt&f <  |!!."R""###D$h$$%%%Z&~& '0'''p((")F)))**8+\++,,, b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d 0@P`p  . < Y " J k X4f}t, bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__ PP)6*644>?IIST^y^yhhr[s[} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__KK 7 7 v vbbNNssY Y""$?%?'')%*%,,. / 1~1~335d6d88:J;J==?.@.BBDEGGIIKcLcNNPEQESSU'V'XXZ [ ]z]z__ahbhddfJgJiik,l,nnpqssuuwaxazz|C}C%%wwWWǐǒ# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSe˃Se˃SSdSSlS!S!S %S %S:S@:SASASmSmSS@SS SUSUSSSS S}S }SSd SCCS CCSWSS WSS\S \SۛSۛS/υS/υSSSSS)S )SSSSpS#lSt#lSySTyS S S߉S$߉SGS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Sz& bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Sz& bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Sz&  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`7Z~2V3Vz /Rv3W+Nr4X b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@{>X|5Ys6Pt- Q  k  .  H l   %Ic&@d~AY}0Tj+Ae;_{Ae<GkRvMqSw*NY}d_ %   e ;!_!!!u""#6###L$p$$ %%%#&G&&&]'' b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d{vqlgb]XSN G @ 9 2 9 +$xW6 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  OO);*;44>!?!IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33##k k""$ % '')*,|,|..01346688:k;k==?Q@QBBD5E5GGIJLLNNPoQoSSUSVSXXZ7[7]]_`bbd e g}g}iikalannpEqEssu)v)xxz { }}ccՄՆGG))   b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_SʃSʃS-Sd-SSlS!S!S%S%SA:S@A:S~S~SsSsS S@ SS SSSSSS SS SSd SICS ICSRS RSC[S C[SSSoȅSoȅS+S+SSS'S 'S#S#SSpS5lSt5lSySTyS/S/S߉S$߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 S& bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 S& bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 S&  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`=wPtCgVzUy%I\Cg+Ob  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@^bf jnrvz ~ $   (   ,   0   4  8 f<Dh@"F$H$o,PMq4X+O<` -xDhLpTx\d lt|$,4 <   b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d^<xV4pN, N   m d  M /  -  m M- m bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  MM)7*744>"?"I`I`ST^^hhrrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``PP @ @  00    l l""$V%V'')@*@,,.*/*11346688:s;s==?]@]BBDGEGGGI2J2LLNOQQSTV@V@XXZ([(]]_`bbd8e8ggi j llnoq|q|ssudvdxxzL{L}}44||ff  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_SY˃SY˃SSdSSlS!S!S%S%S:S@:S5S5SsSsS S@ SS SISISSSS SS SSd SICS ICSRS RS[S [SSSƅSƅSiSiS9S9S7'S 7'S5S5S Sp SGlStGlSySTySSS_S$_SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Se& bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Se& bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Se&  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`+1Uh'K*NTx;_:MqwOsc b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@W{MqCg9]/S%I?5+! x   n  d   Z ~  P t  Fj<`6Z;_0T=a*N?c$HAeBCg<Ei6Gk 0Im*$}wqk e_ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dW.\3 a8f? =    q w J T # 1  |  U.` bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  WW)I*I44>;?;IIS'T'^f^fhiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeWW I I  ;;--  "#%|%|'')n*n,,.`/`113R4R668D9D;;=6>6@@B(C(EEGHJJL M OOQQSoToVVX]Y][[]K^K``b7c7eeg!h!jjl m oJoJqqs0t0vvxy{{}2~2ppTTƑƓ  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_SʃSʃS-Sd-SSlS!S!S%S%SA:S@A:S~S~SsSsS S@ SS SSSSSS SS SSd SICS ICSRS RS\S \SWSWS/ƅS/ƅSSSSS 'S  'SSSSpS)lSt)lSySTyS S S߉S$߉S_S(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 S& bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 S& bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 S&  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`''Kl-Q>bbW{44Xy*: b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@W{MqCg9]/S%I?5+! x   n  d   Z ~  P t  Fj<`.R D6(}o  aVzSw4XEi67[_)M=a?? 1h#xj\ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dW.\3 a8f;   e :  z  d E  9 [c&8 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  KK)5*544>?I[I[S T ^~^~hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``LL 8 8  $$ppZ Z""$D%D'').*.,,./11346w6w88:a;a==?K@KBBD5E5GGIJLLNOQxQxSSU'V'XXZ [ ]]__abdeggiiklllnnpPqPssu4v4xxz{}}nnTTljNj::   b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_S_˃S_˃SSdSSlS!S!S%S%S:S@:S;S;SsSsS S@ SS SOSOSSSS SS SSd SICS ICS7RS 7RSs[S s[SSSDžSDžSeSeSWSWSU'S U'SSSSpS/lSt/lSySTySSS߉S$߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Sd& bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Sd& bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Sd&  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`:d`Brh&J~p(.Rx4  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@SwAe/SA /p ^Lp:^( L   :   ( {   i   W {  Ei3W!E3!tbPt>b,P>,KomMq[OsImQu7[Sw%IUyW{Y} b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dS&yLrEk> d 7   ]  0    kVH)%| bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  QQ)?*?44>-?-IIST^S^Shhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccQQ ? ?  --  n n""$\%\'')J*J,,.8/8113&4&6689;;=>@y@yBBDgEgGGIUJULLN?O?QQS)T)VVXY[[]]_r`rbbdXeXggi>j>llnnp`q`ssuBvBxxz[{[}}==ssUU  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃS3SL3SSTS!S!S%S%SG:S(G:S~S~SsSsS S( SSSSS#S#SS SS SSL SICS ICSRS RS[S [S3S3SSSSS#SD#SSS5lS 5lSMSMSS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 S& bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 S& bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 S&  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`BVz$H(L` Dh2Vjd<`t b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@.R4cEi'K -z\ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1] * w D  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  NN)<*<44>,?,IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``NN < <  **}}k k""$Y%Y'')G*G,,.5/5113#4#6689;;==?v@vBBDdEdGGITJTLLNBOBQQS0T0VVXY[[] ^ ``bbdqeqggi_j_llnHoHqqs,t,vvxy{{}}ff؂؄JJ.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SS˃SS˃SSLSSTS!S!S%S%S:S(:S/S/SySySS(SSSCSCSSSS SS SSL SOCS OCSTS TSM]S M]SSS3S3SSS5SD5S S SGlS GlSSSqS0qSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vB b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 SC* bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vB b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 SC* bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vB  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 SC*  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`*3We#GwVz5Y<Bfyi%I b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@mki g eca_ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 w R -   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  UU)E*E44>5?5IIS!T!^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%% } }""$m%m'')]*],,.M/M113=4=668-9-;;=>@@B C EEGGIuJuLLNaOaQQSMTMVVX9Y9[[]%^%``bceeggisjslln_o_qqsKtKvvx7y7{{}#~#qq]]ӑӓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃS3SL3SSTS!S!S%S%SG:S(G:S~S~SsSsS S( SSSSS#S#SS SS SSL SICS ICSQS QS%[S %[SmSmSÅSÅSSSSDSSS)lS )lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 S& bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 S& bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 S&  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`(w(Lke9]]Uy++OnCg b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@>b D&sUy7[=l bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 ^ +xE } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  JJ)4*444> ? IIS T ^|^|hhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``JJ 4 4  }}ggQ Q""$;%;'')%*%,,./11335n6n88:X;X==?B@BBBD,E,GGIJLLNOQwQwSSUaVaXXZK[K]]_5`5bbdeggi j l~l~nnpbqbssuFvFxxz*{*}}dd։֋HH,, b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SY˃SY˃SSLSSTS!S!S%S%S:S(:S/S/SySySS(SSSCSCSSSS SS SSL SOCS OCSURS URS[S [SٚSٚS%DžS%DžS'S'SSDSSS/lS /lS)S)S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zF b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 SC+ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zF b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 SC+ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zF  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 SC+  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Sw=fTx-Qz(^Ae<hUy,P b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@_][Y}W{UySwQu bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 c >tO } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  OO);*;44>'?'IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''uua a""$M%M'')9*9,,.%/%11346688:s;s==?_@_BBDKEKGGI7J7LLNOQQSTV{V{XXZc[c]]_K`Kbbd3e3ggijllnoqwqwssu_v_xxzG{G}}//ss[[ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS9SL9SSTS!S!S%S%SM:S(M:S~S~SySySS(SSSũSũS)S)SS SS SSL SOCS OCSRS RS9\S 9\SSSSS/S/S/SD/SSSAlS AlSeSeS5S05SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 U( bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 U( bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 U(  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`%I^2V/Shf9]r$Cg|4X% b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@gMq3W=#r X|>b$H bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1j 9  W  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  XX)D*D44>0?0IIST^^histst} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddXX L L  @@44((  "#%%'')r*r,,.^/^113J4J668696;;=">"@@BCEEGGIpJpLLNXOXQQS@T@VVX(Y([[]^``bbdlelggiTjTllnb8\2V,P&J D>8 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 ^ 7  i  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>/?/IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIWJWLLNAOAQQS+T+VVXY[[]]_t`tbbd^e^ggiHjHlln2o2qqstvvxy{{{{}}eeڂڄOOćĉ99## b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS9SL9SSTS!S!S%S%SG:S(G:S~S~SySySS(SSSSS#S#SS SS SSL SOCS OCSQS QS1[S 1[SySyS+ąS+ąS=S=SSDSSS)lS )lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 T' bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 T' bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 T'  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`))Mjo4XX| "]"FcKo b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ .}cIm/S9nTx bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1& u Db } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  TT)<*<44>$?$IIST^z^zhhr^s^} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$ | |""$h%h'')P*P,,.8/8113 4 6689;|;|==?d@dBBDLELGGI4J4LLNOQQSSUnVnXXZR[R]]_6`6bbdeggiikplpnnpTqTssu8v8xxz{}}rrVVȉȋ:: b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSS˃SS˃SSLSSTS!S!S)%S)%S:S(:S)S)SSS#S(#SSS=S=SSSS SS SSL SaCS aCSRS RS[S [SSSƅSƅSSSSDSSS/lS /lS S S߉S0߉SwS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o; b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 SY6 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o; b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 SY6 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o;  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 SY6  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Sw@l g<`@{\)<`|> 0\ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@2,& ysmg bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 B tM& } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP):*:44>$?$IIS T ^}^}hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((vvb b""$M%M'')7*7,,.!/!113 4 6688:j;j==?T@TBBD>E>GGI(J(LLNOQQSSUgVgXXZM[M]]_3`3bbdeggiikrlrnnpXqXssu>v>xxz${$}} }}cc։֋II// b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS?SL?SSTS!S!S5%S5%SS:S(S:S~S~SSS/S(/SSS˩S˩S/S/SS SS S%SL %SmCS mCS+RS +RSg[S g[SSSѽSѽSqSqSSDSSS)lS )lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  k@ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  k@ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  k@  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Sw&J`Rv.Rh\6Zpf>bx  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Y}Ko=a/S!E7)~ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1> h =   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  LL)6*644>?IIST^w^whhr]s]} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%qq[ [""$E%E'')/*/,,./11346x6x88:b;b==?J@JBBD0E0GGIJLLNNPoQoSSUUVUXXZ;[;]]_!`!bbdegzgziik`l`nnpFqFssu,v,xxz{}}kkބކQQÉË55 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSG˃SG˃SSLSSTS!S!S%S%S:S(:SSSSSS(SSS1S1SSSS SS S SL  SUCS UCSQS QS[S [SOSOSSS[S[S/SD/SSSAlS AlSMSMSS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`, b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 SC) bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`, b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 SC) bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`,  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 SC)  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$Tx3W[!Bf!Eb0T-QQuC b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ D.R<`JnX|ft - bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1r U 8  ~  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  YY)E*E44>/?/IIST^^hhrsss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ii]] Q Q  EE99))  "#%w%w'')c*c,,.O/O113;4;668'9';;=>@@BBDoEoGGIWJWLLN?O?QQS'T'VVXY[[]]_k`kbbdSeSggi;j;lln#o#qqs t vvxxzg{g}}OOÂÄ77ww b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS?SL?SSTS!S!S5%S5%SM:S(M:S~S~SSS/S(/SSSũSũS)S)SS SS S%SL %SmCS mCSRS RSO[S O[SSScŅScŅSSSSDSSSlS lS S S]߉S0]߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  j> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  j> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  j>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`\**Ni\/SSw\=X||\ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@p bTxFj8\*N@2 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1g < f; } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  LL)2*244>?IISS]i^ihhrKsK} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''ooU U""$;%;'')!*!,,./1z1z335`6`88:F;F==?*@*BBD E G}G}IIK_L_NNPAQASSU#V#XXZ[]v]v__aXbXddf:g:iiklnnpprosouuwQxQzz|3}3gg׋׍GG b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSG˃SG˃SSLSSTS!S!S%S%S:S(:S#S#SSSS(SSS7S7SSSS SS S SL  SUCS UCSRS RS[S [SESESSɅSSɅSSSSDSSS)lS )lS S S߉S0߉SMS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d0 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 SC* bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d0 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 SC* bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d0  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 SC*  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`CSwr,AeKo/S6ZyA3b;_ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@;%I3WAeOs] k !y!! bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1a D' m } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  SS);*;44>!?!IIST^w^whhr[s[} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggWW G G  77''o o""$W%W'')?*?,,.'/'11346688:k;k==?Q@QBBD5E5GGIJLLNNPoQoSSUSVSXXZ7[7]]_`bbddfqgqiikUlUnnp9q9ssuvxxz{}s}sWWɄɆ;; b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃSESLES ST S!S!S#%S#%SS:S(S:S~S~SSSS(SSS˩S˩S/S/SS SS SSL S[CS [CSESS ESS\S \SɛSɛSۿSۿS S S/SD/SSSAlS AlS5S5SS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 T' bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 T' bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 T'  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Sw'Kb% 3WnY}?cz)$Ko 0] - b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@dHlPtX|`hpx  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1, m M -  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  QQ);*;44>&?&IIST^^hhrlsl} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$p p""$Z%Z'')D*D,,../.11346689;w;w==?a@aBBDKEKGGI6J6LLNOQQSTVzVzXXZb[b]]_J`Jbbd2e2ggijllnoqvqvssu^v^xxzF{F}}..vv`` b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSA˃SA˃SSLS ST S!S!S#%S#%S:S(:SSSSSS(SSS+S+SSSS SS SSL S[CS [CS+RS +RSg[S g[SSS…S…SSSGSDGSSSYlS YlSSSqS0qSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uU! b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 SS' bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uU! b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 SS' bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uU!  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 SS'  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Sw ..Rq [Aee3c3Wv %k b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@88\\66ZZ~~4 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1` N < *   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  [[)M*M44>???IIS+T+^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ii[[ M M  ??11##  "#%%'')r*r,,.d/d113V4V668H9H;;=:>:@@B,C,EEGHJJLMOOQQSsTsVVXaYa[[]O^O``b;c;eeg%h%jjlmooqqshthvvxNyN{{}4~4rrVVȑȓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS?SL?S ST S!S!S#%S#%SS:S(S:S~S~SSSS(SSS˩S˩S/S/SS SS SSL S[CS [CSaRS aRS[S [S嚅S嚅SȅSȅSoSoS#SD#SSS5lS 5lS S S߉S0߉SkS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 T' bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 T' bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 T'  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Sw+-Qhk*NPt6Mqs3'K b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@( 08@$H,P4X< `  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 mM-  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  OO)9*944>#?#IIST^z^zhhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((tt^ ^""$H%H'')2*2,,./11346{6{88:e;e==?O@OBBD9E9GGI#J#LLN O Q|Q|SSUbVbXXZH[H]]_.`.bbdeggiikhlhnnpLqLssu0v0xxz{}}jj܄܆PPÉË66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSA˃SA˃SSLS ST S!S!S#%S#%S:S(:SSSSSS(SSS1S1SSSS SS SSL S[CS [CSsRS sRS[S [SSS̅S̅SSS/SD/SSSAlS AlS/S/S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wY% b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 SS' bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wY% b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 SS' bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wY%  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 SS'  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Sw>hHl2V 0<`Jn0T$Hb<` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@4XX||22 V V!z!! "z"" #0###0$T$ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1tbP> } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  UU)C*C44>1?1IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggUU C C  11  r r""$`%`'')N*N,,.@}@}BBDkEkGGIYJYLLNCOCQQS-T-VVXY[[]^`v`vbbd\e\ggiBjBlln&o&qqstvyvyxxz[{[}}==ssUU b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃSKSLKS ST S!S!S#%S#%SY:S(Y:S~S~SSSS(SSSѩSѩS5S5SS SS SSL S[CS [CSRS RS!\S !\SiSiSSSSS;SD;SSSMlS MlSeSeS5S05SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 T' bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 T' bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 T'  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`SwBPt@Nr-Q>Lp<iJn+ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@2VPtn!?9]W{ u( bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1  r ] H  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  RR)@*@44>0?0IIST^^histst} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddRR @ @  ..  o o""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGIXJXLLNFOFQQS4T4VVX"Y"[[]^``bbdueuggicjcllnLoLqqs0t0vvxy{{}}jj܂܄NN22 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqS5˃S5˃SSLS ST S!S!S#%S#%S:S(:SSSSSS(SSS%S%SSSS SS SSL S[CS [CSQS QS%[S %[SmSmSąSąSSSMSDMS%S%S_lS _lSSSS0S+S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1jJ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 SE& bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1jJ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 SE& bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1jJ  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 SE&  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Sw()MlGk<`d;_46Zx0T . b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@}>Tx+Oe&<`v7 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 N G @ 9 2 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  YY)I*I44>9?9IIS%T%^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiYY I I  99))   ""$q%q'')a*a,,.Q/Q113A4A668191;;=!>!@@BCEEGHJyJyLLNeOeQQSQTQVVX=Y=[[])^)``bceeghjwjwllncocqqsOtOvvx;y;{{}'~'uuaaבד b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃSESLES ST S!S!S#%S#%SY:S(Y:S~S~SSSS(SSSѩSѩS5S5SS SS SSL S[CS [CSRS RS\S \SKSKSʅSʅSGSGS)SD)SSS;lS ;lS S S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 T' bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 T' bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 T'  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Sw&&Jc%IZ~!E &JJnUy b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@"F@d^| /) M G!k!!!e"""# bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 3 t_ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  NN)8*844>$?$IIST^^hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddNN 8 8  ""  kkU U""$?%?''))*),,./11335r6r88:\;\==?F@FBBD0E0GGIJLLNOQ{Q{SSUeVeXXZO[O]]_9`9bbd#e#ggi j llnnpfqfssuJvJxxz.{.}}hhډڋLL00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqS;˃S;˃SSLS ST S!S!S#%S#%S:S(:SSSSSS(SSS+S+SSSS SS SSL S[CS [CSRS RS[[S [[SSS΅S΅SsSsS5SD5S S SGlS GlS;S;S S0 SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1lN b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 SD& bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1lN b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 SD& bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1lN  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 SD&  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Sw5TxCg 0Z~3W6`#G>f  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Nr% I _!!! """6#Z###p$$ %1%%%G&k&&' bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1+$ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  SS)?*?44>+?+IIST^^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++yye e""$Q%Q'')=*=,,.)/)11346689;w;w==?c@cBBDOEOGGI;J;LLN#O#QQS T VVXXZg[g]]_O`Obbd7e7ggijllnoq{q{ssucvcxxzK{K}}33ww__ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSʃSʃSKSLKSSTS!S!S)%S)%S_:S(_:S~S~SSS#S(#SSSשSשS;S;SS SS SSL SaCS aCSRS RS[S [S'S'SÅSÅSKSKSASDASSSSlS SlSwSwSGS0GSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 T' bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 T' bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 T'  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Sw7:^  1}4X5+.R9]%(L~a b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$4Xh&6Zj(8\l* bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1  ~ t j `  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  \\)H*H44>4?4IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hh\\ P P  DD88,,   "#%%'(*v*v,,.b/b113N4N668:9:;;=&>&@@BCEEGGItJtLLN\O\QQSDTDVVX,Y,[[]^``bbdpepggiXjXlln@o@qqs(t(vvxy{{}}llTTȇȉ<<$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwS/˃S/˃SSLSSTS!S!S)%S)%S:S(:S S SSS#S(#SSSSSSSS SS SSL SaCS aCSQS QS[S [S[S[SDžSDžS5S5SGSDGSSSYlS YlS}S}SMS0MSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_?  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 SS' bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_?  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 SS' bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_?   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 SS'  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Sw&&Ju Ae 0[/SAeeA'KKo b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@d 8\0T(Lx Dp< bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1< @ D H L } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  WW)E*E44>3?3IIST^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiWW E E  33!!t t""$b%b'')P*P,,.>/>113,4,6689;;=>@@BBDmEmGGI[J[LLNEOEQQS/T/VVXY[[]^`x`xbbdbebggiLjLlln6o6qqs t vvx y {{}}iiނބSSȇȉ=='' b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSʃSʃSKSLKSSTS!S!S)%S)%S_:S(_:S~S~SSS#S(#SSSשSשS;S;SS SS SSL SaCS aCSRS RS[S [S?S?SͅSͅSSS)SD)SSS;lS ;lS S S߉S0߉SwS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 T' bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 T' bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 T'  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Sw&&Jl(@ds?8Qu~ Ko/S{ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@:^n,< ` p!! ".""">#b###r$$ %0%%% bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1VLB8. } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  XX)@*@44>(?(IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhXX H H  88((  ""$l%l'')T*T,,.>"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwS5˃S5˃SSLSSTS!S!S)%S)%S:S(:SSSSS#S(#SSS%S%SSSS SS SSL SaCS aCSRS RS=[S =[SSSхSхSKSKS/SD/SSSAlS AlSSS߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1aC b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 SS' bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1aC b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 SS' bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1aC  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 SS'  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Sw,<`Gk$>b{;_@dq/SBfn b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@h !4!!!`""#,###X$|$%$%%%P&t&&'''H(l(())) bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1PTX\` } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  TT)>*>44>(?(IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,zzf f""$Q%Q'');*;,,.%/%11346688:n;n==?X@XBBDBEBGGI,J,LLNOQQSSUkVkXXZQ[Q]]_7`7bbdeggijlvlvnnp\q\ssuBvBxxz({(}}ggډڋMM33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S˃S˃SiSiSS$Sw:Sw:S~S~SgS$gS梅S梅SSSSSS$S+RS+RSg[S4g[SçST çSk S k SAS ASS SSlS SlSS S߉S߉SS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1 =k  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1 =k  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1 =k  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 (~rf  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10vv  aa)K*K44>1?1IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvbb N N  ::&&p p""$Z%Z'')D*D,,../.11346689;w;w==?]@]BBDCECGGI)J)LLNOQQSSUhVhXXZN[N]]_4`4bbdeggijlslsnnpYqYssu?v?xxz%{%}} ~~bbԉԋFF** b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S/˃S/˃SS SSS:S:SSSCSCSSpS}S}SSSRS`RS'\S'\SçS$ çSk S k SMS MS%S %S_lS _lSeS eS5S5SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S3 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1[} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S3 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1[} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S3 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1[ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 3WW{ { 1  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10vv  ff)R*R44>:?:IIS"T"^^h i s|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvjj ^ ^  RRFF66""  "#%%'')p*p,,.\/\113H4H668494;;=>@@BCEzEzGGIbJbLLNJOJQQS2T2VVXY[[]^`v`vbbd^e^ggiFjFlln.o.qqstvvxxzr{r}}ZZ΂΄BB''  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}SʃSʃScScSS$Sw:Sw:S~S~SgS$gS梅S梅SSSSSS$SURSURS[S4[SçST çSk S k S)S )SS S;lS ;lS S  S{߉S{߉S#S} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1 >i b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1 >i b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1 >i b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  / / / /  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10vv  __)E*E44>'?'IIS T ^z^zhhrZsZ} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvbb N N  ::&&h h""$N%N'')4*4,,./11346s6s88:Y;Y==?;@;BBDEGGIIKpLpNNPRQRSSU4V4XXZ[]]__aibiddfKgKiik-l-nnpqssuuwbxbzz|D}D&&vvVVƐƒ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S/˃S/˃SS SSS:S:S S SCSCSSpSSSSSSS`SS\S\SçS$ çSk S k S5S 5S S  SGlS GlS S  S߉S߉SkS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S5 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Z} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S5 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Z} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S5 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Z k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 RvDh6Z  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10vv  bb)J*J44>.?.IIST^^hhrfsf} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvff V V  FF66""  ~ ~""$f%f'')N*N,,.6/611346689;z;z==?^@^BBDBEBGGI&J&LLN O Q|Q|SSU`V`XXZD[D]]_(`(bbd e g~g~iikblbnnpFqFssu*v*xxz{}}ddքֆFF((  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃScS cSSSq:Sq:S~S~SaSaS驄Sp驄SMSMSSSRS`RSC[SC[SçS$ çSk S k SGS GSS SYlS YlSSS SS#S#SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 o ]Ko  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10uu  bb)L*L44>5?5IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1uuee U U  EE55""   ""$k%k'')U*U,,.?/?113)4)6689;;==?r@rBBD\E\GGIEJELLN-O-QQSTVVXXZq[q]]_Y`YbbdAeAggi)j)llnoqqssumvmxxzU{U}}==%%mm b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS/˃S/˃SS SSS:S:S S S7S7SSpSSSSSRS`RS[S[SçS$ çSk S k SSS SS+S +SelS elSS SSS+S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\< b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1S} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\< b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1S} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\< b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1S k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 %I5YEiUy  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10rr  dd)V*V44>F?FIIS2T2^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1rrdd V V  HH::,,  "#%%'(*{*{,,.m/m113_4_668Q9Q;;=C>C@@B5C5EEG'H'JJLMOOQRTzTzVVXhYh[[]V^V``bBcBeeg,h,jjlmooqqsotovvxUyU{{};~;!!ww[[͓͑ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃS]S ]SSSq:Sq:S~S~SaSaS驄Sp驄SMSMSSSRS`RS[S[SçS$ çSk S k S5S 5S S  SGlS GlS S  S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 >.RBfVz  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  ^^)H*H44>0?0IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ss__ K K  77##m m""$W%W'')A*A,,.+/+11346688:t;t==?^@^BBDHEHGGI0J0LLNOQQSSUoVoXXZU[U]]_;`;bbdeggijlulunnpYqYssu=v=xxz!{!}}ww[[Ή΋AA'' b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS5˃S5˃SS SSS:S:S S S7S7SSpSSSSSoSS`oSS\S\SçS$ çSk S k SAS ASS SSlS SlSAS ASSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\>  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1S} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\>  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1S} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\>  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1S k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Os;_'K  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10rr  ``)N*N44>:?:IIS"T"^^histst} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1rr`` N N  <<** } }""$k%k'')Y*Y,,.G/G113545668#9#;;=>@@BBDvEvGGIbJbLLNLOLQQS6T6VVX Y [[] ^ ``bbdeeeggiKjKlln/o/qqstvvxxzd{d}}FF(( zz\\ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃSWS WSSSk:Sk:S~S~SgSgS㩄Sp㩄SGSGSSSRS`RS=[S=[SS$ Se S e SGS GSS SYlS YlSqS qSASASS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 |jX|Fj  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  __)M*M44>;?;IIS)T)^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qq__ M M  ;;)) | |""$j%j'')X*X,,.F/F113444668"9";;=>@@BBDuEuGGIcJcLLNQOQQQS?T?VVX-Y-[[]^``b c eegginjnllnWoWqqs;t;vvxy{{}~uuYYˇˉ==!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS5˃S5˃SS SSS:S:SSSSS%Sp%SSSSSyRS`yRS[S[SS$ Se S e SSS SS+S +SelS elSS SSS1S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1eE b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1C} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1eE b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1C} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1eE b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1C k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 >$H.R8\  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  ^^)N*N44>>..  ""$v%v'')f*f,,.V/V113F4F668696;;=&>&@@BCEEGHJ|J|LLNhOhQQSTTTVVX@Y@[[],^,``bceeghjzjzllnfofqqsRtRvvx>y>{{}*~*xxddڑړ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃSWS WSSSe:Se:S~S~SgSgSݩSpݩSASASSSRS`RS[S[SS$ Se S e S5S 5S S  SGlS GlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ;'K7[Gk  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10oo  YY)C*C44>-?-IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ooYY C C  --vv` `""$J%J'')4*4,,./11346}6}88:g;g==?Q@QBBD;E;GGI%J%LLNOQQSSUnVnXXZX[X]]_B`Bbbd,e,ggijllnnpoqossuSvSxxz7{7}}qqUUǎǐ99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS;˃S;˃SS SSS:S:SSSSS%Sp%SSSSSuSS`uSS\S\SS$ Se S e S;S ;SS SMlS MlSAS ASSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1eG b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1C} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1eG b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1C} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1eG b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1C k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 RvDh6Z  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  ZZ)F*F44>0?0IIST^^histst} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nnZZ F F  22  l l""$X%X'')D*D,,.0/011346689;~;~==?j@jBBDVEVGGI@J@LLN(O(QQSTVVXXZl[l]]_T`Tbbd;?;IIS#T#^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qqee Y Y  MMAA55))  "#%%' ( **,,.k/k113W4W668C9C;;=/>/@@BCEEGHJ{J{LLNcOcQQSKTKVVX3Y3[[]^``bcewewggi_j_llnGoGqqs/t/vvxy{{}}ss[[χωCC++ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS;˃S;˃SS SSS:S:SSS7S7S+Sp+SSSSSsRS`sRS[S[SS$ S_ S _ SAS ASS SSlS SlSwS wSGSGSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1nN b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1\} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1nN b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1\} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1nN b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1\ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 : D*N4X  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  XX)F*F44>2?2IIST^^his{s{} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjXX F F  44""u u""$c%c'')Q*Q,,.?/?113-4-6689;;= > @@BBDnEnGGIZJZLLNDODQQS.T.VVXY[[]^`w`wbbdaeaggiKjKlln5o5qqstvvx y {~{~}}hh݂݄RRLJlj<<&& b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃSQS QSSS_:S_:S~S~S[S[SשSpשS;S;SSSRS`RS[S[SS$ S_ S _ S)S )SS S;lS ;lSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1` b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1` b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1` b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 >.RBfVz  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10oo  __)G*G44>-?-IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1oo__ O O  ??//  ""$s%s'')[*[,,.C/C113+4+6689;;==?o@oBBDWEWGGI=J=LLN!O!QQSTVwVwXXZ[[[]]_?`?bbd#e#ggijlylynnp]q]ssuAvAxxz%{%}} {{__щыCC'' b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS;˃S;˃SS SSS:S:SSS7S7S+Sp+SSSSScSS`cSS\S\SS$ S_ S _ S/S /SS SAlS AlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1nP b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1]} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1nP b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1]} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1nP b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1] k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Os<`(L  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  WW)A*A44>)?)IIST^^hhrhsh} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kkWW C C  //}}i i""$T%T'')>*>,,.(/(11346688:q;q==?[@[BBDEEEGGI-J-LLNOQQSSUlVlXXZR[R]]_8`8bbdeggijlwlwnnp]q]ssuCvCxxz){)}}hhۉۋNN44 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSKS KSSS_:S_:S~S~SgSgSשSpשS;S;SSSRS`RS=[S=[SS$ S_ S _ S#S #SS S5lS 5lS S  S߉S߉SkS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Z~RvJnBf  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  UU)?*?44>%?%IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjVV B B  ..zzd d""$N%N'')8*8,,."/"113 4 6688:k;k==?Q@QBBD7E7GGIJLLNOQvQvSSU\V\XXZB[B]]_(`(bbdeggiikglgnnpMqMssu3v3xxz{}}rrVVȉȋ:: b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSA˃SA˃SS SSS:S:SSSSS1Sp1SSSSSQS`QS[S[SS$ S_ S _ S/S /SS SAlS AlSMS MSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wW# b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1D} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wW# b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1D} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wW# b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1D k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 W{Im<`.R  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  ZZ)F*F44>.?.IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jj^^ R R  FF::**  "#%x%x'')d*d,,.P/P113<4<668(9(;;=>@@BBDnEnGGIVJVLLN>O>QQS&T&VVXY[[]]_j`jbbdReRggi:j:lln"o"qqs t v~v~xxzf{f}}NN‚„66tt b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSES ESSSY:SY:S~S~SgSgSѩSpѩS5S5SSSRS`RSU[SU[SS$ S_ S _ SS SS S#lS #lS S  Sc߉Sc߉S S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 SwDhNrX|  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  SS)9*944>?IISS]n^nhhrNsN} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjVV B B  ..vv\ \""$B%B'')(*(,,./11335g6g88:M;M==?/@/BBDEGGIIKdLdNNPFQFSSU(V(XXZ [ ]{]{__a]b]ddf?g?iik!l!nnpqststuuwVxVzz|8}8jjڋڍJJ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSA˃SA˃SS SSS:S:SSSSS1Sp1SSSSSQS`QS[S[SS$ S_ S _ SS SS S/lS /lS S  S߉S߉SMS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wY% b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1D} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wY% b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1D} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wY% b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1D k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 SwAe/SA  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  VV)>*>44>"?"IIST^x^xhhrZsZ} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjZZ J J  ::**r r""$Z%Z'')B*B,,.*/*11346688:n;n==?R@RBBD6E6GGIJLLNNPpQpSSUTVTXXZ8[8]]_`bbdegrgriikVlVnnp:q:ssuvxxz{}t}tXXʄʆ:: b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSES ESSSY:SY:S~S~SUSUSѩSpѩS5S5SSSQS`QS7[S7[SS$ SY S Y S)S )SS S;lS ;lS5S 5SSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 SwAe4X>b  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  VV)@*@44>)?)IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiYY I I  99)) u u""$_%_'')I*I,,.3/311346689;|;|==?f@fBBDPEPGGI9J9LLN!O!QQS T V}V}XXZe[e]]_M`Mbbd5e5ggijllnoqyqyssuavaxxzI{I}}11wwaa b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSG˃SG˃SS SSS:S:SSS+S+S1Sp1SSSSSQS`QS[S[SS$ SY S Y S;S ;SS SMlS MlSS SeSeS S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`, b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1X} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`, b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1X} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`, b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1X k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 SwAe/SA  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  XX)J*J44>:?:IIS&T&^^h i s}s}} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffXX J J  <<..    "#%}%}'')o*o,,.a/a113S4S668E9E;;=7>7@@B)C)EEGHJJL M OOQQSnTnVVX\Y\[[]J^J``b6c6eeg h jjl m o}o}qqsctcvvxIyI{{}/~/kk݌ݎOO b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS?S ?SSSS:SS:S~S~SUSUS˩Sp˩S/S/SSS7RS`7RSs[Ss[SS$ SY S Y SS SS S/lS /lS S  S߉S߉SeS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Sw_ my#  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  RR)<*<44>$?$IIS T ^{^{hhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++wwa a""$K%K'')5*5,,./113 4 6~6~88:h;h==?R@RBBD.?.IIST^^hhrhsh} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffTT B B  00  q q""$_%_'')M*M,,.;/;113)4)6689;;=>@|@|BBDjEjGGIVJVLLN@O@QQS*T*VVXY[[]]_s`sbbdYeYggi?j?lln#o#qqstvvvvxxzX{X}}::nnߎߐPP b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS?S ?SSSM:SM:S~S~S[S[SũSpũS)S)SSSRS`RS=[S=[SS$ SS S S S/S /SS SAlS AlSYS YS)S)SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 SwAeBfLp  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>/?/IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIWJWLLNEOEQQS3T3VVX!Y![[]^``bbdtetggibjbllnMoMqqs1t1vvxy{{}}kk݂݄OO33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSM˃SM˃SS SSS:S:S#S#SSS7Sp7SSSSSQS`QS%[S%[SS$ SS S S SAS ASS SSlS SlSS SwSwSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i5 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1C} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i5 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1C} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i5 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1C k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 SwbPt>b  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  RR)B*B44>0?0IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbRR B B  22"" z z""$j%j'')Z*Z,,.J/J113:4:668*9*;;=>@@B C EEGGIpJpLLN\O\QQSHTHVVX4Y4[[] ^ ``b c eegginjnllndodqqsPtPvvx!?!IIS T ^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccMM 7 7  !!  jjT T""$>%>'')(*(,,./11335q6q88:[;[==?E@EBBD/E/GGIJLLNOQxQxSSUbVbXXZL[L]]_6`6bbd e ggi j llnnpgqgssuKvKxxz/{/}}iiۉۋMM11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSM˃SM˃SS SSS:S:S)S)SSS=Sp=SSSSSRS`RS[S[SS$ SS S S S)S )SS S;lS ;lS/S /S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1k7 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1C} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1k7 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1C} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1k7 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1C k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Sw w --QQu  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  NN):*:44>$?$IIS T ^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&tt` `""$L%L'')8*8,,.$/$11346688:r;r==?^@^BBDJEJGGI4J4LLNOQQSTVxVxXXZ`[`]]_H`Hbbd0e0ggijllnoq|q|ssudvdxxzL{L}}44xx`` b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃS9S 9SSSG:SG:S~S~SISISSpS#S#SSSQS`QSZSZSS SS S| S S5S 5S]~S ]~S;NS ;NSkS kSaމSaމS S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1SwBf3We  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  YY)G*G44>1?1IIST^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeYY M M  AA55))  "#%%'')u*u,,.a/a113M4M668999;;=%>%@@BCEEGGIqJqLLNYOYQQSATAVVX)Y)[[]^``bbdmemggiUjUllnIoIqqs1t1vvxy{{}~uu]]чщEE-- b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17L_SM˃SM˃SS SSS:S:S)S)SSS=Sp=SSSSSQS`QS[S[SS SS S| S S;S ;Sc~S c~SANS ANSqS qSgމSgމSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1r> b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1V} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1r> b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1V} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1r> b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1V k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1SwAeLp"  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10^^  LL)D*D44>0?0IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1^^LL : :  (({{i i""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIXJXLLNBOBQQS,T,VVXY[[]^`u`ubbd_e_ggiIjIlln?o?qqs)t)vvxy{{}}rr\\чщFF00 b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17L_SʃSʃS3S 3SSSG:SG:S~S~SISISSpS#S#SSSRS`RS[S[SS SS S| S SS SE~S E~S#NS #NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1Sw -uMqZ~  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  SS)?*?44>%?%IIS T ^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33## { {""$k%k'')S*S,,.;/;113#4#668 9 ;;==?g@gBBDOEOGGI5J5LLNOQQSSUoVoXXZS[S]]_7`7bbdeggiikqlqnnpaqassuEvExxz){)}} ccՉՋGG++ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17L_SS˃SS˃SS SSS:S:S)S)SSS=Sp=SSSSSqTS`qTS]S]SS SS S| S S#S #SK~S K~S)NS )NSS S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t@ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1W} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t@ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1W} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t@ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1W k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1Sws;_'Bf  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  KK)=*=44>%?%IIS T ^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__KK 7 7  ##qq] ]""$P%P''):*:,,.$/$11346688:m;m==?W@WBBDAEAGGI)J)LLNOQQSSUhVhXXZN[N]]_4`4bbdeggijlslsnnpeqessuKvKxxz1{1}}ppVVɎɐ<< b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations1YSʃSʃSS|SSSSS"S|"S+&SH+&S:S:Si~SPi~SSXSSS%SP%SSS}S}SS$ SS, SS SS$ SS ScDS| cDSSSXSS/]S|/]SwSwSŅSxŅSS\SSS&S&S5S85S S SGlSGlSySySkSlkS;S;SS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1E bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1E  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1E bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Uy&/SnA=Fj ''Kx5ub=a) b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Uy DGk@9]:+O4A .3(%"z{lu ^P t  B f  4 X  & J   <   . ugY}Ko=a/S!E7)~p bTxFj8\*N@ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dU *yRT+)~S(}R' | Q &  {  P  %{P% bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10^^  )D*D44>2?2IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1^^LL : :  ((CC1 1""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIZJZLLNDODQQS.T.VVXY[[]^`w`wbbdaeaggiKjKllnAoAqqs+t+vvxy{{}}tt^^ӇӉHH22# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSe˃Se˃SSdSSlS!S!S%S%S:S@:SASASsSsS S@ SS SUSUSSSS SS SSd SICS ICSIRS IRS[S [S͚S͚S/ƅS/ƅSSSSS 'S  'S)S)SSpS;lSt;lSySTyS/S/S߉S$߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Uu* bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Uu* bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Uu*  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`2Pt2Bft84Xh>&JX| D b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@&JUyGk,P9]2V+O8\A>b3Dh%Jn|nPt `R v  D h  6 Z  ( L   >   0"yoe[QuGk=a3W)MC9 /%|rh^ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d&U*dTC)"~S(}R)  W .   \  3 a8 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  PP)<*<44>'?'IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33##o o""$#%#'') * ,,..01346689;x;x==?b@bBBDLELGGI7J7LLNOQQSTV{V{XXZc[c]]_K`Kbbd?e?ggi'j'llnoqqssukvkxxzS{S}};;##mm  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1YSʃSʃSS|SSSSS"S|"S+&SH+&S9S9Sc~SPc~SSXSSS%SP%SSSwSwSS$ SS, SS SS$ SS ScDS| cDSRSXRSO[S|O[SSSŅSxŅSS\SSS&S&SS8SSS/lS/lSySySSlS߉S߉S}S } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1E bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1E  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1E bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Uy=LpHl$HW{Nr.Ra Tx8\k b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@UyGk9]+OA3%zl ^o P t i  B f c   6 Z ]  ( L W {   > Q u   0Ko"wEii[Mq?c1U#G9+rdVzHl:^,PB b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dU*T)~]S6(}hRA' | Q &  {  P  %zO$ bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  )=*=44>'?'IIS T ^^hhrrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__KK 7 7  ##88$ $""$P%P''):*:,,.$/$11346688:m;m==?W@WBBDAEAGGI+J+LLNOQQSSUjVjXXZP[P]]_6`6bbdeggijlulunnpgqgssuMvMxxz3{3}}rrXXˎː>># b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSk˃Sk˃SSdSSlS!S!S%S%S:S@:SGSGSsSsS S@ SS S[S[SSSS SS SSd SICS ICS3SS 3SSo\S o\SSSq̅Sq̅S-S-SSS(S (SSSSpS/lSt/lSySTyS S S߉S$߉SeS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1` b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Uv* bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1` b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Uv* bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Uv*  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`&J`Bf.Rh6Zp(L>bx l2 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@p(,PTx |48\`Vz @ \   F j b  n h  &   n * N  t  Rvz !z2'6Z^88\\66ZZ~~44XX|| 2  2!V!!!V"z"" #z## $0$$$ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dp`P@0 `?p}`  N  <  *    tbP>, bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  LL):*:44>$?$IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%qq[ [""$%''))+o,o..01346|6|88:f;f==?P@PBBD:E:GGI$J$LLN O Q}Q}SSUcVcXXZI[I]]_/`/bbdeggijlulunnpYqYssu=v=xxz!{!}}ww]]ЉЋCC))  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_SʃSʃS3Sd3SSlS!S!S%S%SG:S@G:S~S~SySySS@SS SSS#S#SS SS SSd SOCS OCSQS QS1[S 1[SySyS3˅S3˅SSSSS(S (S)S)SSpS;lSt;lSySTySASASS$SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 U* bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 U* bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 U*  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$2V{i DmDh 2_2V Rv Ds b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@#Gi!E6C,PAFj?=`;{,9"F< `  V z   p  !   ;  1UKoe2,PJnh93WQuo"@:^X| v )  #!G!!!A"e"""_## b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d#i~TY=4&xaJ3   r ] H 3  t_J bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  RR)>*>44>(?(IdIdST^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbVV J J  >>22""  ""$p%p'')\*\,,.H/H113444668 9 ;;= > @@BBDhEhGGIPJPLLN8O8QQS T V\V\XXZD[D]]_,`,bbdVeVggi>j>lln&o&qqstvvxxzj{j}}RRƂƄ::!!zz  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_S_˃S_˃SSdSSlS!S!S%S%S:S@:S;S;SySySS@SS SOSOSSSS SS SSd SOCS OCS3SS 3SSo\S o\SSSŅSŅSSSSS&S &S)S)SSpS;lSt;lSySTySSSSS#S$#SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Ug* bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Ug* bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Ug*  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`5VzHl*Lp  Bfw(L8\mh, b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@=UyGk?9]A+OCA!E3#G%%I{'Km _Q u  C g  5 Y  ' K   =   /!vhZ~Lp>b0T"F8*qcUyGk9]+OA b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dU*vST0) ~S(}R' | Q &  {  P  %zO$ bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  QQ)?*?44>/?/IIST^Y^Yhisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccQQ ? ?  --  n n""$\%\'')J*J,,.8/8113&4&6689;;=>@y@yBBDgEgGGIWJWLLNEOEQQS3T3VVX!Y![[]^``bbdtetggibjbllnoqqssukvkxxz{}}kk݂݄OO33  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_SʃSʃS3Sd3SSlS!S!S%S%SG:S@G:S~S~SySySS@SS SSS#S#SS SS SSd SOCS OCS-SS -SSi\S i\SSSȅSȅSSSSS'S 'SSSSpS#lSt#lSySTyS S S߉S$߉SAS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 U* bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 U* bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 U*  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` DTx@d*N^ 6Zj D@dt`$ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@b nz$ 0<$H0T<`7H l  6   4 T x   2 `  0 l   . x  ", .*: D(L0T8\@dHlPtX|`hpx ( 0 8  !@! b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dbD&jjLE. {rVT4   t  T  4  tT4 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  NN)6*644>?IUIUST^z^zhhr^s^} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbRR B B  22""j j""$R%R''):*:,,."/"113 4 6~6~88:f;f==?L@LBBD0E0GGIJLLNNPjQjSSUVXXZZ\k]k__abdegtgtiikXlXnnp>   b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_Se˃Se˃SSdSSlS!S!S%S%S:S@:S;S;SySySS@SS SOSOSSSS SS SSd SOCS OCSCRS CRS[S [SǚSǚSŅSŅSSSSS&S &SSSSpS)lSt)lSySTyS S S߉S$߉SqS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Uf* bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Uf* bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Uf*  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`%I^()Mb .-Qf4 1Uj: b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@UyGk9]+OA3%zl ^)MP t + O  B f - Q  4 X / S  & J 1 U   < 3 W   .5Y 7[ugY}Ko=a/S!E7)~p bTxFj8\*N@ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dU*T)~jSG($}R^' | Q &  {  P  %zO$ bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  KK)5*544>!?!IIS T ^H^Hhhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaKK 5 5    ~~hhR R""$<%<'')&*&,,./11335o6o88:Y;Y==?C@CBBD-E-GGIJLLNOQxQxSSUbVbXXZL[L]]_6`6bbd e ggi j llnnp.q.ssuvxxz/{/}}iiۉۋMM11  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃS9SL9SSTS!S!S%S%SM:S(M:S~S~SySySS(SSSũSũS)S)SS SS SSL SOCS OCS1RS 1RSm[S m[SSSSSgSgS5SD5S S SGlS GlSSS_S0_SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 U* bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 U* bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 U*  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`':^l 0,PSw,BEi(47[$ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@B D"F$H&J(L*N bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1yV3 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  TT)F*F44>8?8IIS$T$^^h i s}s}} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbTT F F  88**  "#%y%y'')k*k,,.]/]113O4O668A9A;;=3>3@@B%C%EEGHJJLMO~O~QQSlTlVVXZYZ[[]H^H``b4c4eeghjjlmo{o{qqsatavvxGyG{{}-~-kk݌ݎOO b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SY˃SY˃SSLSSTS!S!S%S%S:S(:S5S5SSSS(SSSISISSSS SS S SL  SUCS UCSSS SS;]S ;]SSSOSOSOSOS/SD/SSSAlS AlSeSeS5S05SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zF b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 UA. bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zF b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 UA. bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zF  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 UA.  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`8\8s\-Q8\ 8\[9 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ 0 .,*(&$"} bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 gBx } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  [[)G*G44>3?3IIST^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1gg[[ O O  CC77++  "#%%'')u*u,,.a/a113M4M668999;;=%>%@@BCEEGGIsJsLLN[O[QQSCTCVVX+Y+[[]^``bbdoeoggiWjWlln?o?qqs't'vvxy{{}}kk߂߄SSLJlj;;## b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_SʃSʃS9SL9SSTS!S!S%S%SG:S(G:S~S~SySySS(SSSSS#S#SS SS SSL SOCS OCSRS RSO[S O[SSSSSSSSDSSS/lS /lSSS߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 U* bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 U* bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 U*  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`!E}Vz'Ku\-Qkb  a3Wh b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@,P. R 0 T 2 V 4 X 6 Z 8\:^ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1mJ'a } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  PP)>*>44>,?,IIST^^hhrhsh} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$[%['')I*I,,.7/7113%4%6689;;=>@x@xBBDfEfGGITJTLLN>O>QQS(T(VVXY[[]]_q`qbbdWeWggi=j=lln!o!qqstvtvtxxzV{V}}88nnߎߐPP b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l_S_˃S_˃SSLSSTS!S!S%S%S:S(:S5S5SSSS(SSSISISSSS SS S SL  SUCS UCSyRS yRS[S [SSSASASSSSDSSS)lS )lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~J b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 UA/ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~J b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 UA/ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~J  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 UA/  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$H\%I$H\2V$H\>b$H\Jn b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ { y  w  u  s  q om bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1S. d? } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  UU)=*=44>%?%IIS T ^{^{hhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%% } }""$i%i'')Q*Q,,.9/9113!4!668 9 ;};}==?e@eBBDMEMGGI5J5LLNOQQSSUoVoXXZS[S]]_7`7bbdeggiikqlqnnpUqUssu9v9xxz{}}ssWWɉɋ;; b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS?SL?SSTS!S!S%S%SS:S(S:S~S~SSSS(SSS˩S˩S/S/SS SS S SL  SUCS UCS3SS 3SSo\S o\SSSSSmSmS;SD;SSSMlS MlSSSqS0qSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 W, bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 W, bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 W,  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`,DhsAe:^f0T\!E&JRva% b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@7=C%I+O1U7[=a bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1rQ0 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  RR)B*B44>2?2IIST^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbRR B B  22"" z z""$j%j'')Z*Z,,.J/J113:4:668*9*;;=>@@B C EEGGIrJrLLN^O^QQSJTJVVX6Y6[[]"^"``bceeggipjplln\o\qqsHtHvvx4y4{{} ~  nnZZБГ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSS˃SS˃SSLSSTS!S!S)%S)%S:S(:S)S)SSS#S(#SSS=S=SSSS SS SSL SaCS aCS=RS =RSy[S y[SSS7S7SCSCSSDSSS)lS )lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o; b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 U\9 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o; b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 U\9 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o;  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 U\9  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`;UybGk%Ip9]1U|+O&=a2 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$~ zvrnj fb bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1Z4h } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  OO)9*944>!?!IIST^z^zhhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((tt^ ^""$H%H'')2*2,,./11346{6{88:e;e==?M@MBBD3E3GGIJLLNNPrQrSSUXVXXXZ>[>]]_$`$bbd e g}g}iikclcnnpIqIssu/v/xxz{}}nnTTƉƋ88 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS?SL?SSTS!S!S%S%SM:S(M:S~S~SSSS(SSSũSũS)S)SS SS S SL  SUCS UCS RS  RSI[S I[SSSŷSŷSSS#SD#SSS5lS 5lS)S)S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 V+ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 V+ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 V+  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`"FX|#$HzZ~&Jl\ (L^^ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@CgI m O s U y [   a gm bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1nM, j } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  NN):*:44>&?&IIST^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&tt` `""$L%L'')8*8,,.$/$11346688:r;r==?^@^BBDJEJGGI6J6LLNOQQSTVzVzXXZb[b]]_J`Jbbd2e2ggijllnoqvqvssu^v^xxzF{F}}..rrZZ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSS˃SS˃SSLSSTS!S!S/%S/%S:S(:S/S/SSS)S()SSSCSCSSS S  SS SSL SgCS gCSmRS mRS[S [SSSSSSSSDSSSlS lS S SW߉S0W߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1s? b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 U[: bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1s? b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 U[: bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1s?  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 U[:  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`#GgZ~CVz .?Rv"F;Nr8\ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@^ Z ~ V z R v N r J n FjBf bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1BvP* } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  MM)3*344>?IISS]j^jhhrLsL} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((ppV V""$<%<'')"*",,./1{1{335a6a88:G;G==?+@+BBD E G~G~IIK`L`NNPBQBSSU$V$XXZ[]w]w__aYbYddf;g;iiklnnpprpspuuwRxRzz|4}4hh؋؍HH b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃSESLES%ST%S!S!S;%S;%SS:S(S:S~S~SSS5S(5SSS˩S˩S/S/SS SS S+SL +SsCS sCSRS RS \S  \SQSQSSSmSmS5SD5S S SGlS GlSkSkS;S0;SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  mB bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  mB bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  mB  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` 1Nr(L HlyLpBfsp<`m& b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ /9C)M3W=aGkQu bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 lM. } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  PP)>*>44>,?,IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$[%['')I*I,,.7/7113%4%6689;;=>@x@xBBDfEfGGITJTLLN>O>QQS(T(VVXY[[]]_q`qbbd[e[ggiEjElln/o/qqstvvxy{x{x}}bbׂׄLL66  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSM˃SM˃SSLS ST S!S!S#%S#%S:S(:S#S#SSSS(SSS7S7SSSS SS SSL S[CS [CSyRS yRS[S [SSS寅S寅SsSsS/SD/SSSAlS AlS5S5SS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d0 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 UA- bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d0 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 UA- bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d0  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 UA-  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`6UyX| 3uZ~)5\%I7_Ei b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$6$H6ZHlZ~l~+ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1eJ/ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  TT)>*>44>)?)IIST^^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggWW G G  77''s s""$]%]'')G*G,,.1/111346689;z;z==?d@dBBDNENGGI9J9LLN!O!QQS T V}V}XXZe[e]]_M`Mbbd5e5ggijllnoqyqyssuavaxxzI{I}}11yycc b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS?SL?SSTS!S!S5%S5%SS:S(S:S~S~SSS/S(/SSS˩S˩S/S/SS SS S%SL %SmCS mCS RS  RSI[S I[SSSSSSSSDSSS/lS /lSSS߉S0߉S}S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  lA bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  lA bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  lA  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`#GZ~$!EyX|CkVzA]Tx b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@[ e  o  y "  , 6@&J bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1pQ2t } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  OO)9*944>#?#IIS T ^|^|hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''uua a""$L%L'')6*6,,. / 113 4 6688:i;i==?S@SBBD=E=GGI'J'LLN O QQSSUfVfXXZL[L]]_2`2bbdeggiikqlqnnpWqWssu=v=xxz#{#}} ||bbՉՋHH.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSM˃SM˃SSLS ST S!S!S#%S#%S:S(:S)S)SSSS(SSS=S=SSSS SS SSL S[CS [CSQS QS[S [SgSgSsSsSSSSDSSS/lS /lS S S߉S0߉SkS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h4 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 UA. bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h4 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 UA. bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h4  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 UA.  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`CUyQuGk9Gk9] 0>b+O&4X} b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ = + O = a O s a s 2 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1y^C(  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  PP):*:44>$?$IIS T ^{^{hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))uu_ _""$I%I'')3*3,,./11346|6|88:f;f==?P@PBBD:E:GGI$J$LLN O Q}Q}SSUcVcXXZI[I]]_/`/bbdeggiikilinnpMqMssu1v1xxz{}}kk݄݆QQĉċ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃSKSLKS ST S!S!S#%S#%SY:S(Y:S~S~SSSS(SSSѩSѩS5S5SS SS SSL S[CS [CSRS RS9\S 9\SSSSSmSmS/SD/SSSAlS AlSMSMSS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 V+ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 V+ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 V+  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Uy4Vz -Tx;_+Rv)Pt!E b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@(2<"F,P6Z@dJn bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1eF' } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  VV)B*B44>,?,IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffZZ N N  BB66&&  ""$t%t'')`*`,,.L/L113848668$9$;;=>@@BBDlElGGITJTLLN3?3IIS!T!^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggUU C C  11  r r""$`%`'')N*N,,.@}@}BBDkEkGGI[J[LLNIOIQQS7T7VVX%Y%[[]^``bcexexggifjfllnOoOqqs3t3vvxy{{}}mm߂߄QQÇÉ55 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃSESLES ST S!S!S#%S#%SY:S(Y:S~S~SSSS(SSSѩSѩS5S5SS SS SSL S[CS [CSRS RSU[S U[SSSSSSSSDSSS/lS /lS S S߉S0߉SMS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 V+ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 V+ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 V+  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Uy>Hl8PtBf2Vz<`,^6Z b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Tx^  h  r  |  % /9C bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1iJ+ m } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  RR):*:44> ? IIST^v^vhhrZsZ} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&&n n""$V%V'')>*>,,.&/&11346688:j;j==?P@PBBD4E4GGIJLLNNPnQnSSURVRXXZ6[6]]_`bbddfpgpiikTlTnnp8q8ssuvxxz{}r}rVVȄȆ:: b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSG˃SG˃SSLS ST S!S!S#%S#%S:S(:S#S#SSSS(SSS7S7SSSS SS SSL S[CS [CSQS QS1[S 1[SySySϹSϹSSSSS)SD)SSS;lS ;lS S S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1y]) b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 UU+ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1y]) b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 UU+ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1y])  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 UU+  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Uy56Z'Mq(L|Ei>n =a2` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@B f j " & J N r v .2VZ~ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations10 p } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  OO)9*944>%?%IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeOO 9 9  ##  llV V""$@%@'')***,,./11335s6s88:];]==?G@GBBD1E1GGIJLLNOQ|Q|SSUfVfXXZP[P]]_:`:bbd$e$ggijllnnpgqgssuKvKxxz/{/}}iiۉۋMM11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃSKSLKSSTS!S!S)%S)%S_:S(_:S~S~SSS#S(#SSSשSשS;S;SS SS SSL SaCS aCSRS RS-\S -\SuSuS'S'SSSGSDGSSSYlS YlSSSqS0qSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 V+ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 V+ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 V+  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Uy -FjzHlu5YJnwuLpy9 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$>4XNrh3)MCg bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1iR;$ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  XX)J*J44>7@@B)C)EEGHJJL M OOQQSpTpVVX^Y^[[]L^L``b8c8eeg"h"jjl m ooqqsetevvxKyK{{}1~1ooSSőœ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqS;˃S;˃SSLSSTS!S!S)%S)%S:S(:SSSSS#S(#SSS+S+SSSS SS SSL SaCS aCSRS RSO[S O[SSS5S5SSSASDASSSSlS SlSwSwSGS0GSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1lN b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 UC* bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1lN b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 UC* bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1lN  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 UC*  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Uy&6ZwMq@dd GkJnp?cTxx b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$>b|?Y}6Zt7Q u  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1{vql } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  __)K*K44>7?7IIST^^his{s{} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kk__ S S  GG;;//##  "#%%'(*y*y,,.e/e113Q4Q668=9=;;=)>)@@BCEEGHJwJwLLN_O_QQSGTGVVX/Y/[[]^``bbdsesggi[j[llnCoCqqs+t+vvxy{{}}ooWWˇˉ??'' b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃSKSLKSSTS!S!S)%S)%S_:S(_:S~S~SSS#S(#SSSשSשS;S;SS SS SSL SaCS aCSRS RS\S \S]S]SSSSS/SD/SSSAlS AlS/S/S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 V+ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 V+ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 V+  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Uy33W))M)MCva9l! b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@]  w  (  B 8 \ Rvl7 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 v_H1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  TT)B*B44>0?0IIST^^hhrlsl} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffTT B B  00  q q""$_%_'')M*M,,.;/;113)4)6689;;=>@|@|BBDjEjGGIXJXLLNBOBQQS,T,VVXY[[]^`u`ubbd[e[ggiAjAlln%o%qqstvxvxxxzZ{Z}}<<rrTT b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSA˃SA˃SSLSSTS!S!S)%S)%S:S(:SSSSS#S(#SSS1S1SSSS SS SSL SaCS aCSQS QS+[S +[SsSsS%S%SSS)SD)SSS;lS ;lS S S߉S0߉SwS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1nR b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 UB* bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1nR b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 UB* bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1nR  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 UB*  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Uy((L{Ko:i?c(W{{4XEii b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@  . R l / Im &Jd'Ae bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1gb]XS  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  YY)A*A44>)?)IIS T ^^hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiYY I I  99))   ""$m%m'')U*U,,.=/=113%4%668 9 ;;==?i@iBBDQEQGGI9J9LLNOQQSTVsVsXXZW[W]]_;`;bbdeggijlulunnpYqYssu=v=xxz!{!}}ww[[͉͋??## b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSʃSʃSQSLQSSTS!S!S/%S/%Se:S(e:S~S~SSS)S()SSSݩSݩSASAS S  SS SSL SgCS gCSRS RS[S [S9S9SϰSϰSSSMSDMS%S%S_lS _lSSSS0S+S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 V+ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 V+ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 V+  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Uy&8\u ?ce< EikDhKoqp b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$0T`(LX| DPt  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1th\P } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  VV)F*F44>6?6IIS"T"^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&& ~ ~""$n%n'')^*^,,.N/N113>4>668.9.;;=>@@BCEEGGIvJvLLNbObQQSNTNVVX:Y:[[]&^&``bceeggitjtlln`o`qqsLtLvvx8y8{{}$~$rr^^ԑԓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwS5˃S5˃SSLSSTS!S!S/%S/%S:S(:SSSSS)S()SSS%S%SSS S  SS SSL SgCS gCSQSS QSS\S \S՛S՛SݱSݱSSS/SD/SSSAlS AlS S S߉S0߉SwS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1aC b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 UU+ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1aC b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 UU+ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1aC  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 UU+  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$,PLp$6Z$Bf,P$Lpl 0 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$Tx"Rv PtNrL p   bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  SS)=*=44>%?%IIS T ^~^~hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,xxb b""$L%L'')6*6,,. / 113 4 6688:i;i==?Q@QBBD7E7GGIJLLNOQvQvSSU\V\XXZB[B]]_(`(bbdeggiikglgnnpMqMssu3v3xxz{}}rrXXʉʋ<<  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSʃSʃSQSLQSSTS!S!S/%S/%Se:S(e:S~S~SSS)S()SSSݩSݩSASAS S  SS SSL SgCS gCSRS RS \S  \SQSQSMSMS}S}S5SD5S S SGlS GlSASASS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 V+ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 V+ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 V+  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Uy((L}>o#G  0a"W{Sww b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@  < H l  x  4 @dp,8\h bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1D8,   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  RR)>*>44>*?*IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **xxd d""$P%P'')<*<,,.(/(11346689;v;v==?b@bBBDNENGGI:J:LLN"O"QQS T V~V~XXZf[f]]_N`Nbbd6e6ggijllnoqzqzssubvbxxzJ{J}}22vv^^ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwS;˃S;˃SSLSSTS!S!S/%S/%S:S(:SSSSS)S()SSS%S%SSS S  SS SSL SgCS gCSSS SSK\S K\SSS{S{SSSSDSSS)lS )lS S So߉S0o߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1cG b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 UU+ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1cG b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 UU+ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1cG  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 UU+  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H~=a  Dzm@v)>r5Y b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ J n  H l FjDhBf bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1$*06  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  QQ)7*744>?IISS]n^nhhrPsP} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,ttZ Z""$@%@'')&*&,,. / 11335e6e88:K;K==?/@/BBDEGGIIKdLdNNPFQFSSU(V(XXZ [ ]{]{__a]b]ddf?g?iik!l!nnpqststuuwVxVzz|8}8ll܋܍LL b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S ˃S ˃SoSoSS$S}:S}:S~S~SgS$gSSSYSYSS$SRSRS[[S4[[SçST çSk S k SYS YS1S 1SklS klSS S_S_SS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19m b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19m b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations19m b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 UyGkLpRv  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ww  ee)S*S44>???IIS)T)^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1wwee S S  AA//   ""$p%p'')^*^,,.L/L113:4:668(9(;;=>@@BCE{E{GGIgJgLLNQOQQQS;T;VVX%Y%[[]^``bbdnenggiXjXllnBoBqqs,t,vvxy{{}~uu__ԇԉII33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S/˃S/˃SS SSS:S:S S SCSCSSpSSSSSRS`RS[S[SçS$ çSk S k SGS GSS SYlS YlSSS SS#S#SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U7 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Y} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U7 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Y} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U7 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Y k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 +OCg[ s  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  aa)K*K44>4?4IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ttdd T T  DD44!!   ""$j%j'')T*T,,.>/>113(4(6689;;==?q@qBBD[E[GGIDJDLLN,O,QQSTVVXXZp[p]]_X`Xbbd@e@ggi(j(llnoqqssulvlxxzT{T}}<<$$ ll b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S˃S˃SiSiSS$S}:S}:S~S~SgS$gSSSYSYSS$S1RS1RSm[S4m[SçST çSk S k SGS GSS SYlS YlS)S )S߉S߉SS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1:k  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1:k  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1:k  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 UyZ~`f   b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10vv  bb)L*L44>4?4IIST^^hissss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvbb N N  ::&&t t""$_%_'')I*I,,.3/311346689;|;|==?f@fBBDPEPGGI8J8LLNOQQSTVwVwXXZ][]]]_C`Cbbd)e)ggijllnnphqhssuNvNxxz4{4}}ssYY̎̐?? b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S5˃S5˃SS SSS:S:S S S=S=SSpSSSSSQSS`QSS\S\SçS$ çSk S k S;S ;SS SMlS MlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U9 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1X} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U9 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1X} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U9 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1X k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Im )Mi -  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  __)I*I44>1?1IIST^^hhrlsl} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1tt`` L L  88$$n n""$X%X'')B*B,,.,/,11346689;u;u==?_@_BBDIEIGGI1J1LLNOQQSSUpVpXXZV[V]]_<`v>xxz"{"}}xx\\ωϋBB(( b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃScS cSSSw:Sw:S~S~SaSaS梅Sp梅SSSSSSSRS`RS[[S[[SçS$ çSk S k SMS MS%S %S_lS _lSeS eS5S5SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 %zl ^  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ww  gg)S*S44>;?;IIS#T#^^h i s}s}} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1wwkk _ _  SSGG77##  "#%%'')q*q,,.]/]113I4I668595;;=>@@BCE{E{GGIcJcLLNKOKQQS3T3VVXY[[]^`w`wbbd_e_ggiGjGlln/o/qqstvvxxzs{s}}[[ςτCC(( b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS5˃S5˃SS SSS:S:SSS7S7S%Sp%SSSSSsRS`sRS[S[SçS$ çSk S k SMS MS%S %S_lS _lSwS wSGSGSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^@  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1U} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^@  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1U} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^@  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1U k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 @"F(L.R  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  ^^)L*L44>:?:IIS(T(^^h i s~s~} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp^^ L L  ::(( { {""$i%i'')W*W,,.E/E113343668!9!;;=>@@BBDtEtGGIbJbLLNPOPQQS>T>VVX,Y,[[]^``bceeggimjmllnVoVqqs:t:vvxy{{}~ttXXʇʉ<<  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃScS cSSSq:Sq:S~S~SaSaS驄Sp驄SMSMSSSRS`RS[S[SçS$ çSk S k S5S 5S S  SGlS GlS S  S߉S߉SeS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 C7[Osh  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10uu  aa)I*I44>-?-IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1uuee U U  EE55!!  } }""$e%e'')M*M,,.5/511346689;y;y==?]@]BBDAEAGGI%J%LLN O Q{Q{SSU_V_XXZC[C]]_'`'bbd e g}g}iikalannpEqEssu)v)xxz { }}ccՄՆEE''  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS;˃S;˃SS SSS:S:SSS7S7S%Sp%SSSSS?SS`?SS{\S{\SçS$ çSk S k S;S ;SS SMlS MlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^B b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1U} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^B b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1U} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^B b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1U k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Fj D\  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  ZZ)D*D44>.?.IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ppZZ D D  ..wwa a""$K%K'')5*5,,./113 4 6~6~88:h;h==?R@RBBDG?GIIS3T3^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ssee W W  II;;--  "#%%'(*|*|,,.n/n113`4`668R9R;;=D>D@@B6C6EEG(H(JJLMOOQRT{T{VVXiYi[[]W^W``bCcCeeg-h-jjlmooqqsptpvvxVyV{{}<~<""xx\\ΑΓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS;˃S;˃SS SSS:S:SSSSS+Sp+SSSSSgRS`gRS[S[SS$ Se S e SGS GSS SYlS YlS}S }SMSMSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1gI b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1A} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1gI b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1A} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1gI b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1A k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 9?!E'K  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  dd)P*P44>:?:IIS"T"^^h i s~s~} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ppdd X X  LL@@44((  "#%%'(*~*~,,.j/j113V4V668B9B;;=.>.@@BCEEGHJzJzLLNbObQQSJTJVVX2Y2[[]^``bcevevggi^j^llnFoFqqs.t.vvxy{{}}rrZZ·ΉBB** b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃS]S ]SSSk:Sk:S~S~SmSmS㩄Sp㩄SGSGSSSRS`RS\S\SS$ Se S e S;S ;SS SMlS MlS;S ;S S SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 (LLpp&  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  __)M*M44>9?9IIS!T!^^hissss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qq__ M M  ;;)) | |""$j%j'')X*X,,.F/F113444668"9";;=>@@BBDuEuGGIaJaLLNKOKQQS5T5VVXY[[] ^ `~`~bbddedggiJjJlln.o.qqstvvxxzc{c}}EE''yy[[ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS;˃S;˃SS SSS:S:SSSSS+Sp+SSSSSKSS`KSS\S\SS$ Se S e S/S /SS SAlS AlS S  S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1gK b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1A} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1gK b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1A} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1gK b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1A k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Im )Mi -  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  ``)H*H44>.?.IIST^^hhrhsh} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp`` P P  @@00    "#%t%t'')\*\,,.D/D113,4,6689;;==?p@pBBDXEXGGI>J>LLN"O"QQSTVxVxXXZ\[\]]_@`@bbd$e$ggijlzlznnp^q^ssuBvBxxz&{&}} ||``҉ҋDD(( b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃSWS WSSSk:Sk:S~S~S[S[S㩄Sp㩄SGSGSSSRS`RSC[SC[SS$ S_ S _ SSS SS+S +SelS elSS SSS1S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 |n `Rv  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10oo  __)O*O44>=?=IIS)T)^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1oo__ O O  ??//  ""$w%w'')g*g,,.W/W113G4G668797;;='>'@@BCEEGHJ}J}LLNiOiQQSUTUVVXAYA[[]-^-``bceeghj{j{llngogqqsStSvvx?y?{{}+~+yyeeۑۓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSA˃SA˃SS SSS:S:SSS7S7S1Sp1SSSSSQS`QS[S[SS$ S_ S _ S)S )SS S;lS ;lS S  S߉S߉SqS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1pR b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1^} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1pR b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1^} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1pR b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1^ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 UyGk9]+O  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  TT)>*>44>$?$IIS T ^}^}hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiUU A A  --yyc c""$M%M'')7*7,,.!/!113 4 6688:j;j==?P@PBBD6E6GGIJLLNOQuQuSSU[V[XXZA[A]]_'`'bbd e ggiikflfnnpLqLssu2v2xxz{}}qqUUljNj99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃSQS QSSSe:Se:S~S~S[S[SݩSpݩSASASSSRS`RS\S\SS$ S_ S _ S;S ;SS SMlS MlSAS ASSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 +OSw {3  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  YY)E*E44>/?/IIST^^hhrsss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mmYY E E  11  k k""$W%W'')C*C,,.///11346689;};}==?i@iBBDUEUGGI?J?LLN'O'QQSTVVXXZk[k]]_S`Sbbd;e;ggi#j#lln o qqssugvgxxzO{O}}77{{cc b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSA˃SA˃SS SSS:S:SSS=S=S1Sp1SSSSSQS`QS[S[SS$ S_ S _ SS SS S)lS )lS S  Si߉Si߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1pT  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1_} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1pT  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1_} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1pT  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1_ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 UyGk9]+O  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  TT):*:44>?IISS]o^ohhrOsO} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kkWW C C  //ww] ]""$C%C''))*),,./11335h6h88:N;N==?0@0BBDEGGIIKeLeNNPGQGSSU)V)XXZ [ ]|]|__a^b^ddf@g@iik"l"nnpqsusuuuwWxWzz|9}9kkۋۍKK b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSQS QSSSe:Se:S~S~SgSgSݩSpݩSASASSSRS`RSC[SC[SS$ S_ S _ S;S ;SS SMlS MlSwS wSGSGSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 {m _Qu  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  YY)G*G44>3?3IIST^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kkYY G G  55##v v""$d%d'')R*R,,.@/@113.4.6689;;= > @@BBDoEoGGI[J[LLNEOEQQS/T/VVXY[[]^`x`xbbdbebggiLjLlln6o6qqs t vvx y {{}}iiނބSSȇȉ=='' b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSG˃SG˃SS SSS:S:SSSSS1Sp1SSSSSQS`QS[S[SS$ S_ S _ S/S /SS SAlS AlS;S ;S S SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1y[' b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1B} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1y[' b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1B} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1y[' b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1B k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 UyHl:^,P  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  UU)?*?44>(?(IIST^^hhrlsl} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhXX H H  88((t t""$^%^'')H*H,,.2/211346689;{;{==?e@eBBDOEOGGI8J8LLN O QQSTV|V|XXZd[d]]_L`Lbbd4e4ggijllnoqxqxssu`v`xxzH{H}}00vv`` b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSKS KSSS_:S_:S~S~SgSgSשSpשS;S;SSSRS`RS \S \SS$ S_ S _ S)S )SS S;lS ;lS S  S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 (LLpp&  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  VV)@*@44>(?(IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjVV B B  ..||h h""$S%S'')=*=,,.'/'11346688:p;p==?Z@ZBBDDEDGGI,J,LLNOQQSSUkVkXXZQ[Q]]_7`7bbdeggijlvlvnnp\q\ssuBvBxxz({(}}ggډڋMM33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSG˃SG˃SS SSS:S:S#S#SSS7Sp7SSSSSQS`QS[S[SS$ S_ S _ S#S #SS S5lS 5lS S  S߉S߉SkS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1y]) b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1B} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1y]) b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1B} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1y]) b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1B k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 UyGk9]+O  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  SS)=*=44>%?%IIS T ^|^|hhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,xxb b""$L%L'')6*6,,. / 113 4 6688:i;i==?S@SBBD=E=GGI%J%LLN O Q~Q~SSUdVdXXZJ[J]]_0`0bbdeggiikjljnnpNqNssu2v2xxz{}}llބކPPÉË66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSKS KSSSY:SY:S~S~SUSUSѩSpѩS5S5SSSRS`RSU[SU[SS$ SY S Y S/S /SS SAlS AlSGS GSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ^^^`  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  [[)G*G44>/?/IIST^^hhrqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kk__ S S  GG;;++  "#%y%y'')e*e,,.Q/Q113=4=668)9);;=>@@BBDoEoGGIWJWLLN?O?QQS'T'VVXY[[]]_k`kbbdSeSggi;j;lln#o#qqs t vvxxzg{g}}OOÂÄ77uu b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSM˃SM˃SS SSS:S:S#S#S+S+S7Sp7SSSSSQS`QS[S[SS$ SY S Y S/S /SS SAlS AlSYS YS)S)SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d0 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Z} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d0 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Z} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d0 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Z k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 UyGk9]+O  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  RR)@*@44>.?.IIST^^hisrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddRR @ @  ..  o o""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGIVJVLLNDODQQS2T2VVX Y [[]^``bbdsesggiajallnJoJqqs.t.vvxy{{}}hhڂڄLL00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSES ESSSY:SY:S~S~SUSUSѩSpѩS5S5SSSRS`RSO[SO[SS$ SY S Y SS SS S)lS )lS S  S߉S߉SGS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 UyLpRvX|  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  UU)=*=44>!?!IIST^w^whhrYsY} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiYY I I  99))q q""$Y%Y'')A*A,,.)/)11346688:m;m==?Q@QBBD5E5GGIJLLNNPoQoSSUSVSXXZ7[7]]_`bbddfqgqiikUlUnnp9q9ssuvxxz{}s}sWWɄɆ99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSM˃SM˃SS SSS:S:S)S)S1S1S=Sp=SSSSSmRS`mRS[S[SS$ SY S Y SS SS S/lS /lSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f2 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1[} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f2 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1[} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f2 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1[ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Uyq#?7[  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  NN)8*844>"?"IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddNN 8 8  ""  kkU U""$?%?''))*),,./11335r6r88:\;\==?F@FBBD0E0GGIJLLNOQyQySSUcVcXXZM[M]]_7`7bbd!e!ggi j llnnpdqdssuHvHxxz,{,}}ff؉؋JJ.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃSES ESSSS:SS:S~S~S[S[S˩Sp˩S/S/SSSQS`QS%[S%[SS$ SS S S S;S ;SS SMlS MlSS SeSeS S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 UyGk9]2V  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  YY)K*K44>;?;IIS'T'^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggYY K K  ==//!!  "#%~%~'')p*p,,.b/b113T4T668F9F;;=8>8@@B*C*EEGHJJL M OOQQSoToVVX]Y][[]K^K``b7c7eeg!h!jjl m ooqqsftfvvxLyL{{}2~2nnRRđē b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSM˃SM˃SS SSS:S:S)S)SSS=Sp=SSSSSQS`QS%[S%[SS$ SS S S S5S 5S S  SGlS GlSkS kS;S;SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m9 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1A} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m9 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1A} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m9 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1A k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 UyZ~Lp>b  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  XX)D*D44>.?.IIST^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddXX L L  @@44((  "#%%'')r*r,,.^/^113J4J668696;;=">"@@BCEEGGInJnLLNVOVQQS>T>VVX&Y&[[]^``bbdjejggiRjRllnDoDqqs,t,vvxy{{}}ppXẊ̉@@(( b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃS?S ?SSSS:SS:S~S~S[S[S˩Sp˩S/S/SSSORS`ORS[S[SS$ SS S S S#S #SS S5lS 5lS#S #S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Uyi} -A  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>-?-IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIUJULLN?O?QQS)T)VVXY[[]]_r`rbbdXeXggi>j>lln&o&qqstvyvyxxz[{[}}==qqSS b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSS˃SS˃SS SSS:S:S)S)SSS=Sp=SSSSS!SS`!SS]\S]\SS$ SS S S SS SS S/lS /lS S  S߉S߉SkS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o; b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1A} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o; b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1A} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o; b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1A k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Uy!MqEi  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  TT)<*<44>"?"IIST^^hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$ | |""$h%h'')P*P,,.8/8113 4 6689;|;|==?d@dBBDLELGGI2J2LLNOQQSSUlVlXXZP[P]]_4`4bbdeggiiknlnnnpZqZssu>v>xxz"{"}}xx\\Ή΋@@$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L_SʃSʃS9S 9SSSM:SM:S~S~SISISũSpũS)S)SSSQS`QSZSZSS SS S| S SAS ASi~S i~SGNS GNSS SމSމSES} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1Uy_Qu`  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  SS)E*E44>3?3IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33## { {""$m%m'')]*],,.M/M113=4=668-9-;;=>@@B C EEGGIsJsLLN_O_QQSKTKVVX7Y7[[]#^#``bceeggiqjqllnioiqqsUtUvvxAyA{{}-~-{{ggݑݓ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17L_SS˃SS˃SS SSS:S:S/S/S%S%SCSpCSSSSSRS`RS \S \SS SS S| S S#S #SK~S K~S)NS )NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vB b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1X} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vB b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1X} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vB b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1X k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1Jn /Uy  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10]]  RR)<*<44>"?"IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1]]II 5 5  !!  wwa a""$K%K'')5*5,,./113 4 6~6~88:h;h==?N@NBBD4E4GGIJLLNOQsQsSSUYVYXXZ?[?]]_%`%bbdeggiikplpnnpVqVssu'?'IIST^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%ss_ _""$O%O'');*;,,.'/'11346688:u;u==?a@aBBDMEMGGI7J7LLNOQQSTV{V{XXZc[c]]_K`Kbbd3e3ggijllnoqqssukvkxxzS{S}};;## gg b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17L_SY˃SY˃SS SSS:S:S/S/S%S%SCSpCSSSSSQS`QS[S[SS SS S| S S S  S3~S 3~SNS NS S  S݉S݉S+S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xD b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Y} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xD b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Y} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xD b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Y k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1'K-Q)M  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  PP)6*644>?IIST^w^whhrWsW} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__KK 7 7  ##ssY Y""$?%?'')%*%,,. / 1~1~335d6d88:J;J==?,@,BBDEGGIIKaLaNNPCQCSSU%V%XXZ[]x]x__afbfddfHgHiik*l*nnp q s}s}uuw_x_zz|A}A##ssSSÐÒ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations1YSʃSʃSS|SSSSS"S|"S+&SH+&S:S:Si~SPi~SSXSSS%SP%SSS}S}SS$ SS, SS SS$ SS ScDS| cDSSSXSS\S|\S5S5SqƅSxqƅS-S\-S!S!S'S'S5S85S S SGlSGlSySySkSlkS;S;SS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1G bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1G  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1G bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`W{&&Jl9 @dg8=a~ Os7 1 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@W{MqCg9]/S%I?5+! x   n  d   Z ~  P t  Fj<`2V=a(L7[B1U8+O .%I$C{=q7k aW{MqCg9]/S%I b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dW.\3 a8f=    q l J C #  |  q  HvM bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  )K*K44>7?7IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaUU I I  ==11eeY Y""$%'(*y*y,,.e/e113Q4Q668=9=;;=)>)@@BCEEGHJwJwLLN_O_QQSGTGVVX/Y/[[]^``bbdsesggi[j[llnOoOqqs7t7vvxy{{}~{{ccׇ׉KK33# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSk˃Sk˃SSdSSlS!S!S%S%S:S@:SGSGSySySS@SS S[S[SSSS SS SSd SOCS OCSORS ORS[S [SӚSӚSkƅSkƅS'S'S!S!S'S 'S;S;SSpSMlStMlSySTySSSeS$eS S(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1` b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Wq. bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1` b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Wq. bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Wq.  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`=f;-Qz=(Ae?<VzA-Q b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@W{MqCg9]/S%I?5+! x   n  d   Z ~  P t  Fj<`:^2V,P(LBB48&{ .m$ {_qQug]SwIm?c5Y+O!E b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dW.\3 a8fj= ?    l i C >   q  HvM bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  VV)J*J44>7@@B)C)EEGHJJL M OOQQSpTpVVX^Y^[[]L^L``b8c8eeg"h"jjl m ooqqsqtqvvxWyW{{}=~=## {{__ёѓ  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1YSʃSʃSS|SSSSS"S|"S+&SH+&S:S:Si~SPi~SSXSSS%SP%SSS}S}SS$ SS, SS SS$ SS ScDS| cDSQSXQS[S|[SaSaSƅSxƅSiS\iS!S!S'S'SS8SSS/lS/lSySyS Sl S߉S߉SkS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1G bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1G  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1G bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`W{&4XmMq&JJnCg<<`9] ..R b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@W{MqCg9]/S%I?5+! x   n  d   Z ~  P t  Fj<`2V(LB8 .${qg 1]+Sw%Im~?cx5Yr+Ol!E f b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dW.\3 a8f=   k B   p U  G .u`L9 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  )A*A44>)?)IIS T ^^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaQQ A A  11!!NN> >""$m%m'')U*U,,.=/=113%4%668 9 ;;==?i@iBBDQEQGGI9J9LLNOQQSTVsVsXXZW[W]]_;`;bbdeggijlulunnpeqessuIvIxxz-{-}}ggىًKK//# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSq˃Sq˃SSdSSlS!S!S%S%S:S@:SGSGSySySS@SS S[S[SSSS SS SSd SOCS OCSRS RS\S \ScScSɅSɅSiSiS#S#S!(S !(S)S)SSpS;lSt;lSySTyS#S#S߉S$߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Wr. bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Wr. bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Wr.  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` --Qp,P?ccZ~ 11Ut#C b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ew$6$H6ZHlZ~l ~  +   =  + O  = a  Osas  .<&J4XBfPtCg^5Y l'K=z% /3!vAh+O9]G k  U!y!!" b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1deJ/y^C( rU 8   ~   a h  D=' gmP bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  PP)B*B44>0?0IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$$%$'')*,,./113)4)6689;;=>@|@|BBDjEjGGIXJXLLNBOBQQS,T,VVXY[[]^`u`ubbd[e[ggiAjAlln1o1qqstvvxxzf{f}}HH** ~~``  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_SʃSʃS9Sd9SSlS!S!S%S%SM:S@M:S~S~SSSS@SS SũSũS)S)SS SS S Sd  SUCS UCSRS RS[[S [[SSSͅSͅSSSSS})S })SSSSpS)lSt)lSySTyS S S߉S$߉S_S(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 X/ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 X/ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 X/  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$Hg"|5Y\'x,Pj t b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@t 0<`l(4Xd ,P\    $ H  T x    @Lp&|8$Dh"} v{4yDhwxu6sFjz8Hl|:Jn~ <  L!p!! """#>###N$r$$ %%%&@& b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dth\PD8, /   ~ g  t B  j   `  VLB8.$ bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10^^  II)3*344>?ITITS T ^~^~hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1^^JJ 6 6  ""nnX X""$B%B''),*,,,./11346u6u88:_;_==?G@GBBD-E-GGIJLLNNPlQlSSUVXXZZ\q]q__abdeggiikglgnnpMqMssu3v3xxz{}}rrXXʉʋ<<   b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_Se˃Se˃SSdSSlS!S!S%S%S:S@:SASASySySS@SS SUSUSSSS SS SSd SOCS OCSSS SS]S ]S_S_SqƅSqƅS-S-S'S'S%'S %'S5S5S Sp SGlStGlSySTySSSkS$kSS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Wi. bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Wi. bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1U  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Wi.  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`(Ld8\itC Hl 0X|Bf@ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@W{MqCg9]/S%I?5+! x   n  d   Z ~  P t  Fj<`o2Ve(L[BQu8Gk .=a$3W{)Mqg]SwIm?c5Y+O!E b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dW.\3 a8fu= L   # k z B Q   p  GuL bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  TT)D*D44>4?4IIS T ^^^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$ | |""$l%l'')\*\,,.L/L113<4<668,9,;;=>@@B C EEGGItJtLLN`O`QQSLTLVVX8Y8[[]$^$``bceeggirjrlln&o&qqstvvxxz{}$~$rr^^ԑԓ  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_SʃSʃS9Sd9SSlS!S!S%S%SG:S@G:S~S~SSSS@SS SSS#S#SS SS S Sd  SUCS UCSSS SS\S \SSS˅S˅SmSmSSS(S (SSSSpSlStlSySTyS S SW߉S$W߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 X/ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 X/ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 X/  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`':^nVz0TW{%'KNrW{ADh&X| b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@m" DBfd;9][ }  2  0 T  R v   t )'KImk >8\Vz t'!Eq?co]m{k .i(L g Fjedc  !5!!!/"S"""M#q##$ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dmZG4!{hUB  -     n x Y SD./ dp bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  II)/*/44>?IIIISS]n^nhhrPsP} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``LL 8 8  $$llR R""$8%8'')*,,./1w1w335]6]88:C;C==?'@'BBD E GzGzIIK\L\NNP>Q>SSUUWVXVZZ\8]8__a]b]ddf?g?iik!l!nnpqststuuwVxVzz|8}8ll܋܍LL  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_Sk˃Sk˃SSdSSlS!S!S%S%S:S@:SASASySySS@SS SUSUSSSS SS SSd SOCS OCSCRS CRS[S [SǚSǚSqƅSqƅS-S-S'S'S%'S %'S#S#SSpS5lSt5lSySTyS)S)S߉S$߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Wh. bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Wh. bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Wh.  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`,,Po /9]] 1))Mj4: b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@W{MqCg9]/S%I?5+! x   n  d   Z ~  P t  Fj<`2V(LB8 .${qCg9] /Sw%Im|?cr5Yh+O^!ETx b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dW.\3 a8f=   k B  (  p   GV-uL bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  NN):*:44>&?&IIST^K^Khhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&tt` `""$L%L'')8*8,,.$/$11346688:r;r==?^@^BBDJEJGGI6J6LLNOQQSTVzVzXXZb[b]]_J`Jbbd2e2ggijllnnp?q?ssu'v'xxzJ{J}}22vv^^  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS?SL?SSTS!S!S%S%SS:S(S:S~S~SSSS(SSS˩S˩S/S/SS SS S SL  SUCS UCSRS RSO[S O[SSSSSSSSDSSS/lS /lS)S)S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 W. bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 W. bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 W.  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$Hd&},PSwsBX| i b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@:^<`>b@dBfDhFjHl bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1<  v S 0  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  NN)8*844>#?#IIS T ^^hhrisi} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaQQ A A  11!!m m""$W%W'')A*A,,.+/+11346688:t;t==?^@^BBDHEHGGI3J3LLNOQQSTVwVwXXZ_[_]]_G`Gbbd/e/ggijllnnpsqsssu[v[xxzC{C}}++ss]] b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS_˃S_˃SSLS ST S!S!S#%S#%S:S(:S5S5SSSS(SSSISISSSS SS SSL S[CS [CSSS SS/]S /]SwSwS;S;SSS/SD/SSSAlS AlSeSeS5S05SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~J b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 W?2 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~J b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 W?2 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~J  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 W?2  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`'Kb 3Wrn*N?cz$Ko 0Rv . b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@#GA;5 /)#| bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 Y 2 d  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  RR)@*@44>.?.IIST^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddRR @ @  ..  o o""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGIVJVLLN@O@QQS*T*VVXY[[]]_s`sbbd]e]ggiGjGlln1o1qqstvvxy{z{z}}ddقلNNÇÉ88"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS?SL?SSTS!S!S%S%SM:S(M:S~S~SSSS(SSSũSũS)S)SS SS S SL  SUCS UCSaRS aRS[S [S嚅S嚅SDžSDžSWSWSSDSSS#lS #lS S S߉S0߉S_S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 W. bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 W. bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 W.  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`,Dhu2>bj@ 8\d(L2V^4X b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@JnLpNrPtRvTxVzX| bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 jG$ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  LL)6*644> ? IIST^w^whhr]s]} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%qq[ [""$E%E'')/*/,,./11346x6x88:b;b==?L@LBBD6E6GGI J LLNOQyQySSU_V_XXZE[E]]_+`+bbdeggiikelennpIqIssu-v-xxz{}}ggلنMM33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS_˃S_˃SSLS ST S!S!S#%S#%S:S(:S;S;SSSS(SSSOSOSSSS SS SSL S[CS [CSmRS mRS[S [SSSƅSƅSSSSDSSS)lS )lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 W?3 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 W?3 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 W?3  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`+-Ql!E4XZ~*N D]4X >b b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@vpj d^X|RvLp bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 = oH! } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  OO)9*944>#?#IIS T ^|^|hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''uua a""$L%L'')6*6,,. / 113 4 6688:i;i==?S@SBBD=E=GGI'J'LLN O QQSSUfVfXXZL[L]]_2`2bbdeggiikqlqnnpWqWssu=v=xxz#{#}} ||bbՉՋHH.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSESLES ST S!S!S#%S#%SS:S(S:S~S~SSSS(SSS˩S˩S/S/SS SS SSL S[CS [CS]SS ]SS\S \SᛅSᛅSSSSS)SD)SSS;lS ;lSYSYS)S0)SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Y0 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Y0 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Y0  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$Ha Hl#GJn 2Fjp:^,P b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@qw}$* 06<B bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1G &  d C  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  OO)=*=44>-?-IIST^^hhrqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaOO = =  ++~~l l""$Z%Z'')H*H,,.6/6113$4$6689;;=>@w@wBBDeEeGGIUJULLNCOCQQS1T1VVXY[[] ^ ``bbdrerggi`j`llnIoIqqs-t-vvxy{{}}ggقلKK// b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSY˃SY˃SSLSSTS!S!S/%S/%S:S(:S/S/SSS)S()SSSCSCSSS S  SS SSL SgCS gCS RS  RSI[S I[SSSSSSS/SD/SSSAlS AlSGSGSS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1s? b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 W^= bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1s? b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 W^= bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1s?  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 W^=  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`#GW{Y}Mq%I[Cg'K_9])M_ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@&* .26:>B bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1@  | Z 8  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  XX)D*D44>.?.IIST^^hhrrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hh\\ P P  DD88((  "#%v%v'')b*b,,.N/N113:4:668&9&;;=>@@BBDnEnGGIVJVLLN>O>QQS&T&VVXY[[]]_j`jbbdReRggi:j:lln"o"qqs t v~v~xxzf{f}}NN‚„66vv b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS?SL?S ST S!S!S#%S#%SS:S(S:S~S~SSSS(SSS˩S˩S/S/SS SS SSL S[CS [CSRS RSO[S O[SSS'ȅS'ȅS{S{SSDSSS/lS /lS S S߉S0߉SwS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 X/ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 X/ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 X/  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` 1Nr""yLp} oJn{eHly b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$H*N0T6Z<`BfHlN r  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1" `? } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  KK)5*544>!?!IIS T ^}^}hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaKK 5 5    ~~hhR R""$<%<'')&*&,,./11335o6o88:Y;Y==?C@CBBD-E-GGIJLLNOQxQxSSUbVbXXZL[L]]_6`6bbd e ggi j llnnpcqcssuGvGxxz+{+}}ee׉׋II-- b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSY˃SY˃SSLSSTS!S!S/%S/%S:S(:S5S5SSS)S()SSSISISSS S  SS SSL SgCS gCSRS RSO[S O[SSSȅSȅSiSiSSDSSS)lS )lS S S߉S0߉SGS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wC b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 W]> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wC b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 W]> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wC  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 W]>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`&^-Qc^&JOs^7Kot^ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@"F&J*N.R2V6Z:^> b  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 tR0 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  RR):*:44> ? IIST^v^vhhrZsZ} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&&n n""$V%V'')>*>,,.&/&11346688:j;j==?P@PBBD4E4GGIJLLNNPnQnSSURVRXXZ6[6]]_`bbddfpgpiikTlTnnp8q8ssuvxxz{}r}rVVȄȆ:: b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃSKSLKS%ST%S!S!S;%S;%SY:S(Y:S~S~SSS5S(5SSSѩSѩS5S5SS SS S+SL +SsCS sCSRS RS9\S 9\SSSSSSS7S7S5SD5S S SGlS GlSkSkS;S0;SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  oF bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  oF bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  oF  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`%%I`3W BHl} .c;_k+O b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@.R8\BfLpVz` jt bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1S 4  v W  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  YY)E*E44>1?1IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeYY M M  AA55))  "#%%'')s*s,,._/_113K4K668797;;=#>#@@BCEEGGIqJqLLNYOYQQSATAVVX)Y)[[]^``bbdmemggiUjUlln=o=qqs%t%vvx y {{}}ii݂݄QQŇʼn99!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSM˃SM˃SSLSSTS!S!S)%S)%S:S(:S)S)SSS#S(#SSS=S=SSSS SS SSL SaCS aCSRS RS[S [S-S-SSSSSASDASSSSlS SlSSSkS0kSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h4 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 W?1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h4 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 W?1 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h4  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 W?1  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`W{9>b7<`=5:^Ei38\q b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@m8.RHlb| -#G bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1.  i R  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  ZZ)L*L44>>?>IIS*T*^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhZZ L L  >>00""  "#%%'')q*q,,.c/c113U4U668G9G;;=9>9@@B+C+EEGHJJL M OOQQSrTrVVX`Y`[[]N^N``b:c:eeg$h$jjlmooqqsgtgvvxMyM{{}3~3qqUUǑǓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃSESLES%ST%S!S!S;%S;%SY:S(Y:S~S~SSS5S(5SSSѩSѩS5S5SS SS S+SL +SsCS sCSRS RS[[S [[SSSȅSȅSSSSDSSS/lS /lS S S߉S0߉SkS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  nE bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  nE bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  nE  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`4Tx+ .W{! 0xY}2n[  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@~' 1;!E+O5Y? c  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations18 z[< } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  UU)=*=44>%?%IIS T ^{^{hhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%% } }""$i%i'')Q*Q,,.9/9113!4!668 9 ;};}==?e@eBBDMEMGGI5J5LLNOQQSSUoVoXXZS[S]]_7`7bbdeggiikqlqnnpUqUssu9v9xxz{}}ssWWɉɋ;; b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSS˃SS˃SSLSSTS!S!S)%S)%S:S(:S/S/SSS#S(#SSSCSCSSSS SS SSL SaCS aCSQS QS+[S +[SsSsSʅSʅSASAS)SD)SSS;lS ;lS)S)S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 W?2 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 W?2 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 W?2  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$X|$HjNr>bbEi88\~;_  0 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@=aW{q"<2 V L!p!!!f"""# bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 ;$ v_ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  TT)B*B44>0?0IIST^^hhrlsl} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffTT B B  00  q q""$_%_'')M*M,,.;/;113)4)6689;;=>@|@|BBDjEjGGIXJXLLNBOBQQS,T,VVXY[[]^`u`ubbd[e[ggiAjAlln%o%qqstvxvxxxzZ{Z}}<<rrTT b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃSKSLKSSTS!S!S)%S)%S_:S(_:S~S~SSS#S(#SSSשSשS;S;SS SS SSL SaCS aCSgRS gRS[S [S뚅S뚅SMSMS7S7S#SD#SSS5lS 5lS S S߉S0߉SkS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 X/ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 X/ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 X/  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`W{&&Jgo1Ud8\`7 0+O b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@'K1U;_EiOsY}c m bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1L -  o P  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  MM)7*744>?IIST^x^xhhr^s^} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&rr\ \""$F%F'')0*0,,./11346y6y88:c;c==?K@KBBD1E1GGIJLLNNPpQpSSUVVVXXZ<[<]]_"`"bbdeg{g{iikalannpGqGssu-v-xxz{}}ll߄߆RRĉċ66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSG˃SG˃SSLSSTS!S!S)%S)%S:S(:S#S#SSS#S(#SSS7S7SSSS SS SSL SaCS aCSORS ORS[S [SӚSӚSąSąSQSQSGSDGSSSYlS YlSSS}S0}S%S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1y]) b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 WW/ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1y]) b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 WW/ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1y])  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 WW/  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`W{,,P&g&J w"Fy#>s6 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@dp,8\h$0T`(L bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1    x l  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  XX)H*H44>8?8IIS$T$^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhXX H H  88((  ""$p%p'')`*`,,.P/P113@4@668090;;= > @@BCEEGHJxJxLLNdOdQQSPTPVVX$H.R8 \  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations11 sT5 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  MM)3*344>?IISS]j^jhhrLsL} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((ppV V""$<%<'')"*",,./1{1{335a6a88:G;G==?+@+BBD E G~G~IIK`L`NNPBQBSSU$V$XXZ[]w]w__aYbYddf;g;iiklnnpprpspuuwRxRzz|4}4hh؋؍HH b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSM˃SM˃SSLSSTS!S!S)%S)%S:S(:S)S)SSS#S(#SSS=S=SSSS SS SSL SaCS aCSRS RSC[S C[SSSͅSͅSSS/SD/SSSAlS AlS5S5SS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{a- b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 WW/ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{a- b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 WW/ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{a-  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 WW/  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`W{&&JwSw4aOsKooLp5YY} b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@X| DP t !!!"<"""H#l##$x$$%4%%% bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1`TH<0 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  RR)>*>44>*?*IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **xxd d""$P%P'')<*<,,.(/(11346689;v;v==?b@bBBDNENGGI:J:LLN"O"QQS T V~V~XXZf[f]]_N`Nbbd6e6ggijllnoqzqzssubvbxxzJ{J}}22vv^^ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃSQSLQSSTS!S!S/%S/%Se:S(e:S~S~SSS)S()SSSݩSݩSASAS S  SS SSL SgCS gCSSS SS?\S ?\SSSSSSS5SD5S S SGlS GlSASASS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 X/ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 X/ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 X/  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`W{&&Jp 0Lp=a(Le(Ko#G b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@/SMqk<6ZTxr% bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1  p [ F  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  RR)<*<44>'?'IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%%q q""$[%['')E*E,,.///11346689;x;x==?b@bBBDLELGGI7J7LLNOQQSTV{V{XXZc[c]]_K`Kbbd3e3ggijllnoqwqwssu_v_xxzG{G}}//wwaa b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSA˃SA˃SSLSSTS!S!S/%S/%S:S(:SSSSS)S()SSS1S1SSS S  SS SSL SgCS gCS RS  RSI[S I[SSSƅSƅSSSASDASSSSlS SlSwSwSGS0GSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1nR b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 WA. bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1nR b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 WA. bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1nR  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 WA.  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`W{&&JSw&&JOs&&JMq&&J b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Cg ,Pr9["Dh -Q bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1    } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  VV)D*D44>2?2IIST^^his{s{} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhVV D D  22  s s""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIZJZLLNDODQQS.T.VVXY[[]^`w`wbbdaeaggiKjKlln5o5qqstvvx y {~{~}}hh݂݄RRLJlj<<&& b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃSQSLQSSTS!S!S/%S/%Se:S(e:S~S~SSS)S()SSSݩSݩSASAS S  SS SSL SgCS gCSRS RS'\S '\SoSoSʅSʅSGSGS)SD)SSS;lS ;lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 X/ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 X/ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 X/  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`W{&8\wBfh1ULprmVz| / b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@C=a[y,& J D!h!!!b"""# bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 1r] } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  PP):*:44>$?$IIS T ^{^{hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))uu_ _""$I%I'')3*3,,./11346|6|88:f;f==?P@PBBD:E:GGI$J$LLN O Q}Q}SSUcVcXXZI[I]]_/`/bbdeggiikilinnpMqMssu1v1xxz{}}kk݄݆QQĉċ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSG˃SG˃SSLSSTS!S!S/%S/%S:S(:SSSSS)S()SSS1S1SSS S  SS SSL SgCS gCSQS QS7[S 7[SSSCЅSCЅSSS/SD/SSSAlS AlSSS߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1pV" b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 W@. bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1pV" b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 W@. bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1pV"  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 W@.  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`W{&&JOs&&JGk&&J?c&&J b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@s : \!!!#"""E#i## $$$.%R%%%t&&';'''](( bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  SS)=*=44>'?'IIS T ^^hhrfsf} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++yye e""$P%P''):*:,,.$/$11346688:m;m==?W@WBBDAEAGGI+J+LLNOQQSSUjVjXXZP[P]]_6`6bbdeggijlulunnp[q[ssuAvAxxz'{'}} ffىًLL22 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSʃSʃSWSLWSSTS!S!S5%S5%Sk:S(k:S~S~SSS/S(/SSS㩄S㩄SGSGSS SS S%SL %SmCS mCSRS RS[S [SESESÅSÅSKSKSASDASSSSlS SlSkSkS;S0;SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 X/ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 X/ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 X/  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`W{&&JyBgC:UyOs2Cg  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$4Xh&6Zj(8\l* bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1  ~ t j `  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>1?1IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIYJYLLNGOGQQS5T5VVX#Y#[[]^``bbdvevggidjdllnMoMqqs1t1vvxy{{}}kk݂݄OO33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwS;˃S;˃SSLSSTS!S!S5%S5%S:S(:SSSSS/S(/SSS+S+SSSS SS S%SL %SmCS mCS?SS ?SS{\S {\SÛSÛSDžSDžSqSqSASDASSSSlS SlSYSYS)S0)SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1cG b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 WW/ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1cG b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 WW/ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1cG  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 WW/  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$HIm<<` 0TTx#GFjl_! b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@l<p@t Dx$H|(L bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1d l t |  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  \\)H*H44>2?2IIST^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ll`` T T  HH<<,,  "#%z%z'')f*f,,.R/R113>4>668*9*;;=>@@BBDrErGGIZJZLLNBOBQQS*T*VVXY[[]]_n`nbbdVeVggi>j>lln&o&qqstvvxxzj{j}}RRƂƄ::!!zz b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSʃSʃSWSLWSSTS!S!S5%S5%Se:S(e:S~S~SSS/S(/SSSݩSݩSASASS SS S%SL %SmCS mCSRS RS\S \ScScSͅSͅSSS/SD/SSSAlS AlSSS߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 X/ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 X/ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 X/  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`W{&+O~5Yv+O?cncImm# b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@:^n,< ` p!! ".""">#b###r$$ %0%%% bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1VLB8. } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  OO)9*944>%?%IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeOO 9 9  ##  llV V""$@%@'')***,,./11335s6s88:];]==?G@GBBD1E1GGIJLLNOQ|Q|SSUfVfXXZP[P]]_:`:bbd$e$ggijllnnpgqgssuKvKxxz/{/}}iiۉۋMM11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSA˃SA˃SSLSSTS!S!S5%S5%S:S(:SSSSS/S(/SSS+S+SSSS SS S%SL %SmCS mCS]SS ]SS\S \SᛅSᛅS}҅S}҅SSS)SD)SSS;lS ;lS S S߉S0߉SYS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1eK b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 WW/ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1eK b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 WW/ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1eK  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 WW/  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$HNr(:^,Pt4X 0fx> b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ ,!P!!!""0#T##$$$4%X%%&&&8'\''(((<)`)) *** bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 $, } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  VV)>*>44>$?$IIST^z^zhhr^s^} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjZZ J J  ::**r r""$Z%Z'')B*B,,.*/*11346688:n;n==?T@TBBD8E8GGIJLLNOQrQrSSUVVVXXZ:[:]]_`bbdegtgtiikXlXnnp>  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S ˃S ˃SoSoSS$S:S:S~S~SgS$gSSS_S_SS$SRSRS[[S4[[SɧST ɧSq S q SYS YS1S 1SklS klSS S_S_SS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations15o b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations15o b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations15o b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 W{MqNrPt  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10zz  nn)Z*Z44>D?DIIS,T,^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1zznn b b  VVJJ>>22  "&#&%%'(**,,.t/t113`4`668L9L;;=8>8@@B$C$EEGHJJLLNlOlQQSTTTVVXI?IIIS5T5^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1uugg Y Y  KK==//!!  "#%%'(*~*~,,.p/p113b4b668T9T;;=F>F@@B8C8EEG*H*JJLMOOQRT}T}VVXkYk[[]Y^Y``bEcEeeg/h/jjlmooqqsrtrvvxXyX{{}>~>$$zz^^БГ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S ˃S ˃SoSoSS$S}:S}:S~S~SgS$gSSSYSYSS$S[RS[RS[S4[SɧST ɧSq S q SAS ASS SSlS SlSS S߉S߉SS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations16m b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations16m b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations16m b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 W{gw#3  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10xx  hh)P*P44>6?6IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1xxhh X X  HH88((  "#%|%|'')d*d,,.L/L1134446689;;=>@x@xBBD`E`GGIFJFLLN*O*QQSTVVXXZd[d]]_H`Hbbd,e,ggijllnnpfqfssuJvJxxz.{.}}hhډڋLL00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S;˃S;˃SS SSS:S:SSS=S=S%Sp%SSSSS'SS`'SSc\Sc\SɧS$ ɧSq S q SGS GSS SYlS YlSGS GSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W=  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1V} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W=  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1V} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1W=  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1V k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 @dt2Bf  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10uu  cc)Q*Q44>=?=IIS%T%^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1uucc Q Q  ??--   ""$n%n'')\*\,,.J/J113848668&9&;;=>@@BCEyEyGGIeJeLLNOOOQQS9T9VVX#Y#[[] ^ ``bbdhehggiNjNlln2o2qqstvvxxzg{g}}II++ }}__ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S˃S˃SiS iSSS}:S}:S~S~SgSgSSpSYSYSSS7RS`7RSs[Ss[SçS$ çSk S k S;S ;SS SMlS MlS S  S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  -%}u  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  ^^)H*H44>.?.IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ss__ K K  77##m m""$W%W'')A*A,,.+/+11346688:t;t==?Z@ZBBD@E@GGI&J&LLN O QQSSUeVeXXZK[K]]_1`1bbdeggiikplpnnpVqVssu???IIS+T+^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qqaa Q Q  AA11!!  "#%y%y'')i*i,,.Y/Y113I4I668999;;=)>)@@BCEEG H JJLLNkOkQQSWTWVVXCYC[[]/^/``bceeghj}j}llnioiqqsUtUvvxAyA{{}-~-{{ggݑݓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S˃S˃SiS iSSSw:Sw:S~S~SgSgS梅Sp梅SSSSSSSIRS`IRS[S[SçS$ çSk S k S#S #SS S5lS 5lS S  S{߉S{߉S#S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  0,($  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  \\)B*B44>$?$IIST^w^whhrWsW} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ss__ K K  77##  e e""$K%K'')1*1,,./11335p6p88:V;V==?8@8BBDEGGIIKmLmNNPOQOSSU1V1XXZ[]]__afbfddfHgHiik*l*nnp q s}s}uuw_x_zz|A}A##ssSSÐÒ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S;˃S;˃SS SSS:S:SSS=S=S+Sp+SSSSSSS`SSQ\SQ\SçS$ çSk S k SGS GSS SYlS YlSMS MSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`F b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1W} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`F b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1W} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`F b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1W k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 =am)5Y  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  ]])I*I44>3?3IIST^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qq]] I I  55!!  o o""$[%['')G*G,,.3/31134668 9 ;;==?m@mBBDYEYGGICJCLLN+O+QQSTVVXXZo[o]]_W`Wbbd?e?ggi'j'llnoqqssukvkxxzS{S}};;## gg b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃScS cSSSw:Sw:S~S~SmSmS梅Sp梅SSSSSSS+RS`+RSg[Sg[SS$ Se S e SAS ASS SSlS SlSMS MSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 )vl   b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10rr  __)I*I44>2?2IIST^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1rrbb R R  BB22  ~ ~""$h%h'')R*R,,.e>ggi&j&llnoqqssujvjxxzR{R}}::"" jj b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSA˃SA˃SS SSS:S:SSSSS1Sp1SSSSS[RS`[RS[S[SS$ Se S e SGS GSS SYlS YlS}S }SMSMSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1iM b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1?} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1iM b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1?} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1iM b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1? k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 468:  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  [[)I*I44>5?5IIST^^h i s~s~} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mm[[ I I  77%% x x""$f%f'')T*T,,.B/B1130406689;;= > @@BBDqEqGGI]J]LLNGOGQQS1T1VVXY[[]^`z`zbbddedggiNjNlln8o8qqs"t"vvx y {{}}kkUUʇʉ??)) b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃS]S ]SSSq:Sq:S~S~SmSmS驄Sp驄SMSMSSSRS`RS'\S'\SS$ Se S e S5S 5S S  SGlS GlSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 -QUy }5  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  [[)E*E44>-?-IIST^^hhrhsh} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp\\ H H  44    j j""$T%T'')>*>,,.(/(11346688:q;q==?[@[BBDEEEGGI-J-LLNOQQSSUlVlXXZR[R]]_8`8bbdeggijlrlrnnpVqVssu:v:xxz{}}ttXXˉˋ>>$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSA˃SA˃SS SSS:S:SSSSS1Sp1SSSSS!SS`!SS]\S]\SS$ Se S e S5S 5S S  SGlS GlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1iO b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1?} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1iO b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1?} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1iO b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1? k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 @dt2Bf  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  ZZ)D*D44>,?,IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nnZZ F F  22  l l""$W%W'')A*A,,.+/+11346688:t;t==?^@^BBDHEHGGI0J0LLNOQQSSUoVoXXZU[U]]_;`;bbd!e!ggijlzlznnp`q`ssuFvFxxz,{,}}kkމދQQĎĐ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃS]S ]SSSq:Sq:S~S~S[S[S驄Sp驄SMSMSSSRS`RS[[S[[SS$ Se S e SAS ASS SSlS SlSkS kS;S;SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 "yoe  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  \\)J*J44>8?8IIS&T&^^h i s|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nn\\ J J  88&& y y""$g%g'')U*U,,.C/C1131416689;;= > @@BBDrErGGI`J`LLNNONQQS1?1IIST^^hissss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mmaa U U  II==--  "#%{%{'')g*g,,.S/S113?4?668+9+;;=>@@BBDqEqGGIYJYLLNAOAQQS)T)VVXY[[]]_m`mbbdUeUggi=j=lln%o%qqs t vvxxzi{i}}QQłń99ww b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃSWS WSSSk:Sk:S~S~S[S[S㩄Sp㩄SGSGSSS-SS`-SSi\Si\SS$ Se S e S/S /SS SAlS AlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 <`x:Rv  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  VV)@*@44>*?*IIST^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1llVV @ @  **ss] ]""$G%G'')1*1,,./11346z6z88:d;d==?N@NBBD8E8GGI"J"LLN O QQSSUkVkXXZU[U]]_?`?bbd)e)ggijllnnplqlssuPvPxxz4{4}}nnRRĎĐ66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSG˃SG˃SS SSS:S:S#S#S=S=S7Sp7SSSSSQS`QS1[S1[SS$ Se S e S#S #SS S5lS 5lS S  S߉S߉SYS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1rX$ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1a} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1rX$ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1a} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1rX$ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1a k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 W{MqCg9]  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  YY)A*A44>%?%IIS T ^{^{hhr]s]} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mm]] M M  ==-- u u""$]%]'')E*E,,.-/-11346688:q;q==?U@UBBD9E9GGIJLLNOQsQsSSUWVWXXZ;[;]]_`bbdeguguiikYlYnnp=q=ssu!v!xxz{}w}w[[̈́͆== b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃSWS WSSSe:Se:S~S~SgSgSݩSpݩSASASSSRS`RSU[SU[SS$ S_ S _ SAS ASS SSlS SlSwS wSGSGSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ~tk a  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  bb)N*N44>8?8IIS T ^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nnbb V V  JJ>>22&&  "#%%'(*|*|,,.h/h113T4T668@9@;;=,>,@@BCEEGHJxJxLLN`O`QQSHTHVVX0Y0[[]^``bcetetggi\j\llnDoDqqs,t,vvxy{{}}ppXẊ̉@@(( b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSM˃SM˃SS SSS:S:S#S#SSS7Sp7SSSSSQS`QS+[S+[SS$ S_ S _ SAS ASS SSlS SlSS SkSkSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{_+ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1@} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{_+ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1@} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{_+ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1@ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 W{NrDh:^  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  [[)M*M44>=?=IIS)T)^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ii[[ M M  ??11##  "#%%'')r*r,,.d/d113V4V668H9H;;=:>:@@B,C,EEGHJJL M OOQQSqTqVVX_Y_[[]M^M``b9c9eeg#h#jjl m ooqqsftfvvxLyL{{}2~2nnRRđē b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃSQS QSSSe:Se:S~S~SgSgSݩSpݩSASASSSSS`SSW\SW\SS$ S_ S _ S)S )SS S;lS ;lS S  S߉S߉SwS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 :^r2Fj  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  \\)D*D44>*?*IIST^^hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ll\\ L L  <<,,   ""$p%p'')X*X,,.@/@113(4(6689;;==?l@lBBDTETGGI:J:LLNOQQSTVtVtXXZX[X]]_<`v>xxz"{"}}xx\\Ή΋@@$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSM˃SM˃SS SSS:S:S)S)SSS=Sp=SSSSSyRS`yRS[S[SS$ S_ S _ S/S /SS SAlS AlS/S /S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{a- b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1@} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{a- b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1@} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{a- b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1@ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 W{ u'C;_  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  WW)E*E44>1?1IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiWW E E  33!!t t""$b%b'')P*P,,.>/>113,4,6689;;=>@@BBDmEmGGIYJYLLNCOCQQS-T-VVXY[[]^`v`vbbd\e\ggiBjBlln&o&qqstvyvyxxz[{[}}==qqSS b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSQS QSSS_:S_:S~S~SUSUSשSpשS;S;SSS[RS`[RS[S[SS$ SY S Y SS SS S/lS /lS S  S߉S߉SkS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1` b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1` b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1` b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 gy&8&J  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  RR)<*<44>"?"IIST^{^{hhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++wwa a""$K%K'')5*5,,./113 4 6~6~88:h;h==?N@NBBD4E4GGIJLLNOQsQsSSUYVYXXZ?[?]]_%`%bbd e g~g~iikdldnnpJqJssu0v0xxz{}}ooSSʼnŋ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSM˃SM˃SS SSS:S:S)S)S1S1S=Sp=SSSSSQS`QS%[S%[SS$ SY S Y SAS ASS SSlS SlSS SwSwSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h4 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1\} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h4 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1\} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h4 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1\ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 W{MqCg9]  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  UU)E*E44>3?3IIST^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%% } }""$m%m'')]*],,.M/M113=4=668-9-;;=>@@B C EEGGIsJsLLN_O_QQSKTKVVX7Y7[[]#^#``bceeggiqjqlln]o]qqsItIvvx5y5{{}!~! oo[[ёѓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSKS KSSS_:S_:S~S~SUSUSשSpשS;S;SSSQS`QS1[S1[SS$ SY S Y S S  SS SlS lS S  S]߉S]߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 W{MqCg>b  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  PP)6*644>?IISS]k^khhrKsK} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++ssY Y""$?%?'')%*%,,. / 1~1~335d6d88:J;J==?,@,BBDEGGIIKaLaNNPCQCSSU%V%XXZ[]x]x__aZbZddf'?'IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))wwc c""$O%O'');*;,,.'/'11346688:u;u==?a@aBBDMEMGGI7J7LLNOQQSTV{V{XXZc[c]]_K`Kbbd3e3ggijllnoqwqwssu_v_xxzG{G}}//ss[[ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃSES ESSSY:SY:S~S~SaSaSѩSpѩS5S5SSSRS`RS[[S[[SS$ SY S Y S)S )SS S;lS ;lS/S /S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ^bf k  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  SS)=*=44>&?&IIST^^hhrlsl} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&&r r""$\%\'')F*F,,.0/011346689;y;y==?c@cBBDMEMGGI6J6LLNOQQSTVzVzXXZb[b]]_J`Jbbd4e4ggijllnoqxqxssu`v`xxzH{H}}00vv`` b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSS˃SS˃SS SSS:S:S/S/S S SCSpCSSSSSQS`QS7[S7[SS$ SY S Y S5S 5S S  SGlS GlSkS kS;S;SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q= b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1?} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q= b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1?} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q= b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1? k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 W{_UyKo  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  OO)=*=44>)?)IIST^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaOO = =  ++~~l l""$Z%Z'')H*H,,.6/6113$4$6689;;=>@w@wBBDeEeGGIQJQLLN;O;QQS%T%VVXY[[]]_n`nbbdXeXggiBjBlln6o6qqs t vvx y {{}}iiނބSSȇȉ=='' b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃSES ESSSS:SS:S~S~S[S[S˩Sp˩S/S/SSSRS`RSC[SC[SS$ SY S Y SS SS S/lS /lS S  S߉S߉SeS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 W{PtRvTx  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  OO)9*944>!?!IIS T ^|^|hhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((tt^ ^""$H%H'')2*2,,./11346{6{88:e;e==?O@OBBD9E9GGI!J!LLNOQzQzSSU`V`XXZF[F]]_,`,bbdeggiikjljnnpNqNssu2v2xxz{}}llބކPPÉË66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSY˃SY˃SS SSS:S:S/S/S S SCSpCSSSSSSS`SS\S\SS$ SY S Y S#S #SS S5lS 5lSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1s? b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1?} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1s? b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1?} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1s? b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1? k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 W{ 1y/S Qu  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  NN)8*844> ? IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&tt` `""$K%K'')5*5,,./113 4 6~6~88:h;h==?R@RBBD.?.IIST^^h i s~s~} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGIVJVLLNDODQQS2T2VVX Y [[]^``bbdsesggiajallnVoVqqs:t:vvxy{{}~ttXXʇʉ<<  b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LeSY˃SY˃SS SSS:S:S5S5S%S%SISpISSSSSRS`RS[S[SS SS S| S S5S 5S]~S ]~S;NS ;NSMS MSCމSCމSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zF b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Z} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zF b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Z} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zF b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Z k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations18\(L`  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  QQ)G*G44>/?/IIS#T#^^h i s}s}} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaUU I I  ==11!!   ""$y%y'')e*e,,.Q/Q113=4=668)9);;=>@@BBDoEoGGIWJWLLN?O?QQS'T'VVXY[[]]_k`kbbd_e_ggiGjGlln/o/qqstvvxxzs{s}}[[ςτCC(( b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LeSʃSʃS?S ?SSSM:SM:S~S~SOSOSũSpũS)S)SSSQS`QS[S[SS SS S| S S#S #SK~S K~S)NS )NSS S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1W{j w  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  JJ)8*844>"?"IIS T ^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``JJ 4 4  }}ggQ Q""$?%?''))*),,./11335r6r88:\;\==?F@FBBD0E0GGIJLLNOQyQySSUcVcXXZM[M]]_7`7bbd!e!ggi j llnnppqpssuTvTxxz8{8}}rrVVȎȐ:: b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LeSY˃SY˃SS SSS:S:S5S5S%S%SISpISSSSSQS`QS[S[SS SS S| S SS SE~S E~S#NS #NS S  S݉S݉SsS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|H b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1[} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|H b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1[} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|H b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1[ k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1"F$H>  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  MM)=*=44>!?!IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaQQ A A  11!!  i i""$Y%Y'')A*A,,.)/)11346688:m;m==?Q@QBBD5E5GGIJLLNNPoQoSSUSVSXXZ7[7]]_`bbd e g}g}iikalannpEqEssu)v)xxz { }}ccՄՆEE''  b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations1YSʃSʃSS|SSSSS"S|"S1&SH1&S :S :So~SPo~SSXSSS+SP+SSSSSS$ SS, S S  SS$ S!S !SiDS| iDSSSXSSW\S|W\SSSƅSxƅSS\SKSKSI'SI'SAS8ASSSSlSSlSySySSlSwSwSS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1I bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1I  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1I bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Y}>Hl% DNrQu&JTx,PZ~Im b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@>Y}Sw<Mq:Gk:Ae8;_65Y4/S2)M# G   A   ;   5   /  )#|vpj d^X|RvLpFj@d:^4X.R(L"F@ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dY2u Pd+=oH!zS  ,   ^ 7   iBt bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  )I*I44>9?9IIS%T%^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__OO ? ?  //KK; ;""$q%q'')a*a,,.Q/Q113A4A668191;;=!>!@@BCEEGHJyJyLLNeOeQQSQTQVVX=Y=[[])^)``bceeghjwjwllnoooqqs[t[vvxGyG{{}3~3 mm# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSq˃Sq˃SSdSSlS!S!S%S%S:S@:SMSMSSSS@SS SaSaSSSS SS S Sd  SUCS UCSRS RS\S \SKSKSƅSƅSSSKSKSI'S I'S/S/SSpSAlStAlSySTySYSYS)S$)SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Ym2 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Ym2 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Ym2  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` 00Tw3W*MqqW{%IImC|2 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@:Y},SwMqsGkeAeW{;_Im5Y;_/S-Q)M# G   A   ;   5   /  )#|vpj d^X|RvLpFj@d:^4X.R(L"F@ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dYk2@ dj=oH!zS  ,   ^ 7   iBt bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  PP)@*@44>0?0IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$!%!'')*,,..013'4'6689;;=>@z@zBBDhEhGGIXJXLLNFOFQQS4T4VVX"Y"[[]^``bbdueuggicjcllnXoXqqs>""  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1YSʃSʃSS|SSSSS"S|"S1&SH1&S:S:Si~SPi~SSXSSS+SP+SSS}S}SS$ SS, S S  SS$ S!S !SiDS| iDS RSX RSI[S|I[SSSƅSxƅSS\SKSKSI'SI'S)S8)SSS;lS;lSySyS/Sl/S߉S߉SS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1I bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1I  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1I bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`]$$Hi_;_ba 16Zvc 0 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Y}SwMqGkAe;_5Y/S)M 0# G .   A  ,   ;  *   5  (   /  &  )$#"|}vpj d^X|RvLpFj@d:^4X.R(L"F@ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dY2 d=ao<H!rzMS  ,   ^ 7   iBt bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  )?*?44>+?+IIST^^his{s{} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__KK 7 7  ##44 ""$Q%Q'')=*=,,.)/)11346689;w;w==?c@cBBDOEOGGI;J;LLN#O#QQS T VVXXZg[g]]_O`Obbd7e7ggijllnoqqssuovoxxzW{W}}??''kk# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xYSw˃Sw˃SSdSSlS!S!S%S%S:S@:SMSMSSSS@SS SaSaSSSS SS S Sd  SUCS UCSRS RS[S [SSS DžS DžSSSWSWSU'S U'SSSSpS/lSt/lSySTySSS߉S$߉S}S(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Yn2 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Yn2 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Yn2  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`+Bfm<,PW{0T:AeHl$+Ou` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Z~VzRvNrJnFjBf>b:^C6 Z  5  2 V  '  } . R   p * N b  & J T x  "FFjB=7 1+%~xrl f`Z~TxNrHlBf<` b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dZ4hB?viP>*]  6   h A   sL%~ bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  JJ)8*844>$?$IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``JJ 4 4  }}ggQ Q""$%'w'w))+a,a..01335r6r88:\;\==?F@FBBD0E0GGIJLLNOQ{Q{SSUeVeXXZO[O]]_9`9bbd#e#ggi j llnoqrqrssuVvVxxz:{:}}ttXXʎʐ<<  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_SʃSʃS?Sd?SSlS!S!S%S%SS:S@S:S~S~SSSS@SS S˩S˩S/S/SS SS S Sd  SUCS UCSSS SS\S \S;S;SƅSƅSSSQSQSO'S O'S/S/SSpSAlStAlSySTySeSeS5S$5SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Z3 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Z3 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Z3  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Y}?Jn;ALpCNrQu!EPt3 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Y}=7Uy 1Os -ImCg'=a!z7[t1Un+O% I   C   =   7   1  +%~xrl f`Z~TxNrHlBf<`6Z0T*N$HB b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dY2r Kd$=}oH!zS  ,   ^ 7   iBt bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  MM);*;44>)?)IISS]^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__MM ; ;  ))||j j""$X%X'')F*F,,.4/4113"4"6689;;==?u@uBBDcEcGGIQJQLLN;O;QQS%T%VVXY[[]]_n`nbbdeggijl{l{n6o6qqs t vvx y {{}}iiނބSSȇȉ==''  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_Sk˃Sk˃SSdSSlS!S!S%S%S:S@:SASASSSS@SS SUSUSSSS SS S Sd  SUCS UCS)TS )TSe]S e]SSSDžSDžSSSQSQSO'S O'S)S)SSpS;lSt;lSySTySeSeS5S$5SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Yk2 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Yk2 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Yk2  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` //St%|Fjj<`<<` 2|<`  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Y} 1'Sw~MqtImj Cg`=aVz7[Lp1UBf+O% I   C   =   7   1  +%~xrl f`Z~TxNrHlBf<`6Z0T.R,P*N b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dY d2; di=oH!zS  ,   ^ 7   iBt bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  ZZ)F*F44>2?2IIST^T^Thisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffZZ N N  BB66**  "#%%'')t*t,,.`/`113L4L668898;;=$>$@@BCEEGGIrJrLLNZOZQQSBTBVVX*Y*[[]^``bbdnenggiVjVllnoqxqxssu`v`xxz{}}llTTȇȉ<<$$  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_SʃSʃS?Sd?SSlS!S!S%S%SM:S@M:S~S~SSSS@SS SũSũS)S)SS SS S Sd  SUCS UCSRS RSO[S O[SSSƅSƅSSSQSQSO'S O'SSSSpS/lSt/lSySTyS S S߉S$߉SwS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Z3 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Z3 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Z3  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Y}&&Jh]4X[$a)Mfe  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Y}SwMqGkAe;_5Y/S+Oh% I   b  C   \  =  V z  7  P t  1  J n + Dh%~>bxrl f`Z~TxNrHlBf<`6Z0T*N$HB b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dY2 d=Vo/H!az:U  .   ` 9   kDv bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  LL)6*644> ? IISS]^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``LL 8 8  $$rr^ ^""$I%I'')3*3,,./11346|6|88:f;f==?P@PBBD:E:GGI$J$LLN O Q}Q}SSUcVcXXZI[I]]_/`/bebeddfKgKiik1l1nnp\q\ssuBvBxxz({(}}ggډڋMM33  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_Sk˃Sk˃SSdSSlS!S!S%S%S:S@:SGSGSSSS@SS S[S[SSSS SS S Sd  SUCS UCS=RS =RSy[S y[SSSƅSƅSSSQSQSO'S O'SSSSpS)lSt)lSySTyS S S߉S$߉SeS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Yj2 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Yj2 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Yj2  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` .Hlv86Zd6$HRv46@d2 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Y}SwMqGkAe;_5Y/S)M8\# G . R   A $ H   ;  >   5  4   /  *  ) w#|vpj d^X|RvLpFj@d:^4X.R(L"F@ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dY2 d=@qnJE#|U  .   ` 9   kDv bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  TT)<*<44>$?$IIST^A^Ahhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$ | |""$h%h'')P*P,,.8/8113 4 6689;|;|==?d@dBBDLELGGI4J4LLNOQQSSUnVnXXZR[R]]_6`6bbdeggiikplpnnpqssuuwqxqz { }}vvZZ̉̋>>""  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSESLES ST S!S!S#%S#%SS:S(S:S~S~SSSS(SSS˩S˩S/S/SS SS SSL S[CS [CSRS RS[[S [[SSSSSISIS#SD#SSS5lS 5lSASASS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Y2 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Y2 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Y2  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`@~Lp>xJn<rHl:pFj b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@: 0&}si_UyKo bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1mDr } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  SS)?*?44>)?)IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccWW K K  ??33##  ""$q%q'')]*],,.I/I113545668!9!;;= > @@BBDiEiGGIQJQLLN9O9QQS!T!VVX Y [}[}]]_e`ebbdMeMggi5j5llnoqqstvyvyxxza{a}}II11qq b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSe˃Se˃SSLSSTS!S!S)%S)%S:S(:S;S;SSS#S(#SSSOSOSSSS SS SSL SaCS aCSQS QS1[S 1[SySyS+S+SCSCSSDSSS)lS )lS S S߉S0߉S_S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Y=6 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Y=6 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Y=6  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`*Y}*NmSw;__Nr --QrIm! b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@("{uoi c]W{ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1]6h } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  NN)8*844> ? IIST^y^yhhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''ss] ]""$G%G'')1*1,,./11346z6z88:d;d==?L@LBBD2E2GGIJLLNNPqQqSSUWVWXXZ=[=]]_#`#bbd e g|g|iikblbnnpHqHssu.v.xxz{}}mmSSʼnŋ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃS?SL?S ST S!S!S#%S#%SS:S(S:S~S~SSSS(SSS˩S˩S/S/SS SS SSL S[CS [CSRS RS[S [S?S?S϶S϶SSSSDSSS#lS #lS S S߉S0߉SAS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Y2 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Y2 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Y2  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$Hc#G)MPtEi<Rv|g b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Ae7 [ - Q # G  =  3 ) bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1I wN% } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  OO)7*744>?IIST^s^shhrWsW} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33##k k""$S%S'');*;,,.#/#113 4 6688:g;g==?M@MBBD1E1GGIJLLNNPkQkSSUOVOXXZ3[3]]_`bbddfmgmiikQlQnnp5q5ssuvxxzz|o}oSSńņ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSe˃Se˃SSLSSTS!S!S)%S)%S:S(:SASASSS#S(#SSSUSUSSSS SS SSL SaCS aCSRS RS\S \ScScSSSSSSDSSSlS lS S SW߉S0W߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Y=7 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Y=7 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Y=7  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` 1vNr4BfsDh4Xex&JW{6 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@QuK o E i ? c 9 ] 3 W -Q'K bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1AsL% } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  LL)2*244>?IISS]i^ihhrKsK} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''ooU U""$;%;'')!*!,,./1z1z335`6`88:F;F==?*@*BBD E G}G}IIK_L_NNPAQASSU#V#XXZ[]v]v__aXbXddf:g:iiklnnpprosouuwQxQzz|3}3gg׋׍GG b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSKSLKSSTS!S!S)%S)%SY:S(Y:S~S~SSS#S(#SSSѩSѩS5S5SS SS SSL SaCS aCSRS RS9\S 9\SSSSS=S=S;SD;SSSMlS MlSSSeS0eS S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 [4 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 [4 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 [4  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`ANr5Y;Hl5eBf! /<`.R b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ -tfX|Jn<`/S!E bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 ^3] } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  UU)G*G44>9?9IIS%T%^^h i s~s~} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccUU G G  99++  "#%z%z'')l*l,,.^/^113P4P668B9B;;=4>4@@B&C&EEGHJJLMOOQQSmTmVVX[Y[[[]I^I``b5c5eeghjjl m o|o|qqsbtbvvxHyH{{}.~.llތގPP‘“ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSY˃SY˃SSLSSTS!S!S5%S5%S:S(:S5S5SSS/S(/SSSISISSSS SS S%SL %SmCS mCS7RS 7RSs[S s[SSSSS[S[S)SD)SSS;lS ;lS5S5SS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wC b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Y`A bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wC b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Y`A bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wC  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Y`A  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$Z~$Hl"gDhhs@@d<) b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$,4< D(L0T8\ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1`@  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  SS)=*=44>(?(IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&&r r""$\%\'')F*F,,.0/011346689;y;y==?c@cBBDMEMGGI8J8LLN O QQSTV|V|XXZd[d]]_L`Lbbd4e4ggijllnoqxqxssu`v`xxzH{H}}00xxbb b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSESLESSTS!S!S)%S)%SY:S(Y:S~S~SSS#S(#SSSѩSѩS5S5SS SS SSL SaCS aCSRS RS[[S [[SSSSScScS#SD#SSS5lS 5lS#S#S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Z3 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Z3 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Z3  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H`! DzMq7tAevn b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@7 ) ~  p b T x F j 8\ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations12\1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  QQ)?*?44>-?-IIST^^hhrisi} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccQQ ? ?  --  n n""$\%\'')J*J,,.8/8113&4&6689;;=>@y@yBBDgEgGGIUJULLN?O?QQS)T)VVXY[[]]_r`rbbdXeXggi>j>lln"o"qqstvuvuxxzW{W}}99ooQQ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS_˃S_˃SSLSSTS!S!S5%S5%S:S(:S;S;SSS/S(/SSSOSOSSSS SS S%SL %SmCS mCSQS QS7[S 7[SSS׷S׷SSSSDSSS/lS /lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{G b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Y_B bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{G b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Y_B bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{G  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Y_B  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`,Y}DhsSw:^fMq0T\Gk&JRv b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@dH l P t X | `  h px  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1`@ ` } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  OO)9*944>#?#IIST^z^zhhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((tt^ ^""$H%H'')2*2,,./11346{6{88:e;e==?O@OBBD9E9GGI#J#LLN O Q|Q|SSUbVbXXZH[H]]_.`.bbdeggiikhlhnnpLqLssu0v0xxz{}}jj܄܆PPÉË66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃSKSLKS+ST+S!S!SA%SA%S_:S(_:S~S~SSS;S(;SSSשSשS;S;SS SS S1SL 1SyCS yCS-SS -SSi\S i\SSSSS=S=SASDASSSSlS SlSSSwS0wSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  qJ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  qJ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  qJ  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`8<`>b /3W%z)M|<CrTx b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@${qg]SwIm?c5Y bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1W.\ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  SS)C*C44>3?3IIST^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33## { {""$k%k'')[*[,,.K/K113;4;668+9+;;=>@@B C EEGGIsJsLLN_O_QQSKTKVVX7Y7[[]#^#``bceeggiqjqlln]o]qqsItIvvx5y5{{}!~! oo[[ёѓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSS˃SS˃SSLSSTS!S!S/%S/%S:S(:S/S/SSS)S()SSSCSCSSS S  SS SSL SgCS gCSRS RS\S \SKSKS?S?SSS5SD5S S SGlS GlS_S_S/S0/SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Y=5 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Y=5 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Y=5  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Y}&&Jy8g$H&UyyX|Cgg b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$B<`Z~ x+%ICga bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1kVA, } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  TT)B*B44>2?2IIS T ^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffTT B B  00  q q""$_%_'')M*M,,.;/;113)4)6689;;=>@|@|BBDjEjGGIZJZLLNHOHQQS6T6VVX$Y$[[]^``bcewewggiejellnNoNqqs2t2vvxy{{}}llނބPP‡‰44 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃSKSLKS+ST+S!S!SA%SA%S_:S(_:S~S~SSS;S(;SSSשSשS;S;SS SS S1SL 1SyCS yCSIRS IRS[S [S͚S͚SöSöSoSoS)SD)SSS;lS ;lS/S/S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  pI bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  pI bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  pI  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$Hg5"5Yj5'Bfj5>5 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@+O! E  ; 1  ' ~  t j  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations13 a8 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  OO);*;44>'?'IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''uua a""$M%M'')9*9,,.%/%11346688:s;s==?_@_BBDKEKGGI7J7LLNOQQSTV{V{XXZc[c]]_K`Kbbd3e3ggijllnoqwqwssu_v_xxzG{G}}//ss[[ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSY˃SY˃SSLSSTS!S!S/%S/%S:S(:S5S5SSS)S()SSSISISSS S  SS SSL SgCS gCSRS RS=[S =[SSS-S-S#S#S#SD#SSS5lS 5lSSS߉S0߉S}S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p< b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Y=6 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p< b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Y=6 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p<  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Y=6  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Y}'6ZqUy0TTxQu*NNrMq$HHl b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@   2 , P J n h 93WQu bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1mXC } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  NN)8*844>$?$IIST^^hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddNN 8 8  ""  kkU U""$?%?''))*),,./11335r6r88:\;\==?F@FBBD0E0GGIJLLNOQ{Q{SSUeVeXXZO[O]]_9`9bbd#e#ggi j llnnpfqfssuJvJxxz.{.}}hhډڋLL00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃSQSLQSSTS!S!S/%S/%Se:S(e:S~S~SSS)S()SSSݩSݩSASAS S  SS SSL SgCS gCSRS RS!\S !\SiSiSSSmSmS;SD;SSSMlS MlSqSqSAS0ASS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Z3 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Z3 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Z3  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` --Q/S  DuW{6g(Y}}7 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$ 0<$H0T<`HlTx bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1bD& } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  QQ)?*?44>-?-IIST^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccQQ ? ?  --  n n""$\%\'')J*J,,.8/8113&4&6689;;=>@y@yBBDgEgGGIUJULLN?O?QQS)T)VVXY[[]]_r`rbbd\e\ggiFjFlln0o0qqstvvxy{y{y}}cc؂؄MM‡‰77!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSM˃SM˃SSLSSTS!S!S/%S/%S:S(:S)S)SSS)S()SSS=S=SSS S  SS SSL SgCS gCS[RS [RS[S [SߚSߚS簅S簅SSS;SD;SSSMlS MlSwSwSGS0GSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{a- b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 YY3 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{a- b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 YY3 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{a-  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 YY3  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Y}&&J*m,P ..R 122V!E b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$4Xh&6Zj(8\l *  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1vlbX } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  ^^)J*J44>6?6IIST^^hiszsz} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jj^^ R R  FF::..""  "#%%'(*x*x,,.d/d113P4P668<9<;;=(>(@@BCEEGHJvJvLLN^O^QQSFTFVVX.Y.[[]^``bbdrerggiZjZllnBoBqqs*t*vvxy{{}}nnVVʇʉ>>&& b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃSQSLQSSTS!S!S/%S/%Se:S(e:S~S~SSS)S()SSSݩSݩSASAS S  SS SSL SgCS gCS=RS =RSy[S y[SSSSSSS)SD)SSS;lS ;lS S S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Z3 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Z3 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Z3  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$Hp22Pt .1Um*5W{& b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@` l  x " .  : "F.R:^ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1jL.r } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP):*:44>$?$IIS T ^}^}hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((vvb b""$M%M'')7*7,,.!/!113 4 6688:j;j==?T@TBBD>E>GGI(J(LLNOQQSSUgVgXXZM[M]]_3`3bbdeggiikrlrnnpXqXssu>v>xxz${$}} }}cc։֋II// b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSS˃SS˃SSLSSTS!S!S/%S/%S:S(:S)S)SSS)S()SSS=S=SSS S  SS SSL SgCS gCSRS RSO[S O[SSSSSSS#SD#SSS5lS 5lS S S߉S0߉SwS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}e1 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 YY3 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}e1 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 YY3 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}e1  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 YY3  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Y}&,P{Y}*Nm Y}(L_Y}&JQu b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ : ^ n  , < ` p .>br 0 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1ND:0&  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  XX)@*@44>(?(IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhXX H H  88((  ""$l%l'')T*T,,.>"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃSWSLWSSTS!S!S5%S5%Sk:S(k:S~S~SSS/S(/SSS㩄S㩄SGSGSS SS S%SL %SmCS mCSSS SSQ\S Q\SSScScSSS;SD;SSSMlS MlSSSSS#S0#SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Z3 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Z3 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Z3  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Y}&&J{!<n Ei /` !Rvv1U b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$"FDhf=;_] bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1mZG4 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  WW)C*C44>-?-IIST^^hhrqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1gg[[ O O  CC77''  ""$u%u'')a*a,,.M/M113949668%9%;;=>@@BBDmEmGGIUJULLN=O=QQS%T%VVX Y [[]]_i`ibbdQeQggi9j9lln!o!qqs t v}v}xxze{e}}MM55uu b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSG˃SG˃SSLSSTS!S!S5%S5%S:S(:S#S#SSS/S(/SSS7S7SSSS SS S%SL %SmCS mCSSS SSW\S W\SSSSSSS)SD)SSS;lS ;lS S S߉S0߉SwS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1pV" b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Y?2 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1pV" b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Y?2 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1pV"  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Y?2  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H?c>>bs4XX| 1 NrrBf# b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$Jn9](Lr;a *  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  RR)<*<44>$?$IIS T ^}^}hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++wwa a""$K%K'')5*5,,./113 4 6~6~88:h;h==?P@PBBD6E6GGIJLLNOQuQuSSU[V[XXZA[A]]_'`'bbd e ggiikflfnnpLqLssu2v2xxz{}}qqWWɉɋ;; b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃSWSLWSSTS!S!S5%S5%Se:S(e:S~S~SSS/S(/SSSݩSݩSASASS SS S%SL %SmCS mCSRS RS[S [S-S-SuSuSASAS#SD#SSS5lS 5lS S S߉S0߉SSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Z3 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Z3 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Z3  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`[$$H~  Dz;@x?c>tg b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ 4 2 V T x v  + )MKom" bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1!{hU } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  SS);*;44>!?!IIST^w^whhr[s[} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggWW G G  77''o o""$W%W'')?*?,,.'/'11346688:k;k==?Q@QBBD5E5GGIJLLNNPoQoSSUSVSXXZ7[7]]_`bbddfqgqiikUlUnnp9q9ssuvxxz{}s}sWWɄɆ;; b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSM˃SM˃SSLSSTS!S!S5%S5%S:S(:S#S#SSS/S(/SSS7S7SSSS SS S%SL %SmCS mCSORS ORS[S [SӚSӚS߻S߻SSSSDSSS)lS )lS S Si߉S0i߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1rZ& b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Y>2 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1rZ& b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Y>2 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1rZ&  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Y>2  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H;,,P944X7<<`5 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@P t  ? c  .RxAg 0Vz bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  PP)6*644>?IISS]m^mhhrOsO} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++ssY Y""$?%?'')%*%,,. / 1~1~335d6d88:J;J==?.@.BBDEGGIIKcLcNNPEQESSU'V'XXZ [ ]z]z__a\b\ddf>g>iik l nnpqssssuuwUxUzz|7}7kkۋۍKK b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSʃSʃS]SL]SSTS!S!S5%S5%Sq:S(q:S~S~SSS/S(/SSS驄S驄SMSMSS SS S%SL %SmCS mCSRS RS\S \SWSWS S SSSMSDMS%S%S_lS _lSSSwS0wSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 Z3 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 Z3 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 Z3  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`Y}&&J**N'K ..R[44X b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$8\p 0Dh|<Pt$ H  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1xph` } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  YY)K*K44>=?=IIS)T)^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggYY K K  ==//!!  "#%~%~'')p*p,,.b/b113T4T668F9F;;=8>8@@B*C*EEGHJJL M OOQQSqTqVVX_Y_[[]M^M``b9c9eeg#h#jjl m ooqqsftfvvxLyL{{}2~2ppTTƑƓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSA˃SA˃SSLSSTS!S!S5%S5%S:S(:SSSSS/S(/SSS+S+SSSS SS S%SL %SmCS mCS3SS 3SSo\S o\SSSSSSS;SD;SSSMlS MlSGSGSS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1eK b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 YY3 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1eK b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 YY3 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1eK  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 YY3  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$HFj0TTx~`>@Rv(LLp b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$\ .f8pBz(L 2 V  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 ( } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  WW)A*A44>,?,IIST^^hhrrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjZZ J J  ::** v v""$`%`'')J*J,,.4/411346689;};}==?g@gBBDQEQGGI1?1IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggUU C C  11  r r""$`%`'')N*N,,.@}@}BBDkEkGGIYJYLLNCOCQQS-T-VVXY[[]^`v`vbbd\e\ggiBjBlln&o&qqstvyvyxxz[{[}}==ssUU b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSA˃SA˃SSLSSTS!S!S5%S5%S:S(:SSSSS/S(/SSS1S1SSSS SS S%SL %SmCS mCSQSS QSS\S \S՛S՛SSSISIS/SD/SSSAlS AlSSS߉S0߉S}S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1gO b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 YY3 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1gO b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 YY3 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1gO  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 YY3  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$HKoBBf<``+OZ~~  1k / b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@  < `  FjPt"Z~,d6 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations12<FPZ  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  SS)=*=44>'?'IIS T ^~^~hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,xxb b""$L%L'')6*6,,. / 113 4 6688:i;i==?S@SBBD=E=GGI'J'LLN O QQSSUfVfXXZL[L]]_2`2bbdeggiiklllnnpPqPssu4v4xxz{}}nnTTljNj::  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S˃S˃SuSuSS$S:S:S~S~SgS$gSSSeSeSS$SRSRSU[S4U[SɧST ɧSq S q SeS eS=S =SwlS wlSS SSSIS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations14q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations14q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations14q b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Y}SwPtNr  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10xx  hh)X*X44>F?FIIS2T2^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1xxhh X X  HH88((  "#%%'')p*p,,.`/`113P4P668@9@;;=0>0@@B C EEGHJJLLNrOrQQS^T^VVXJYJ[[]6^6``b"c"eeghjjllnpopqqs\t\vvxHyH{{}4~4  nn b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S;˃S;˃SS SSS:S:SSS=S=S+Sp+SSSSSgRS`gRS[S[SɧS$ ɧSq S q SSS SS+S +SelS elS}S }SMSMSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y?  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1U} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y?  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1U} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Y?  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1U k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 <=?A  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  aa)O*O44>=?=IIS+T+^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ssaa O O  ==++ ~ ~""$l%l'')Z*Z,,.H/H113646668$9$;;=>@@BCEwEwGGIeJeLLNSOSQQSATAVVX/Y/[[]^``b c eeggipjpllnYoYqqs=t=vvx!y!{{}~ww[[͇͉??## b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S˃S˃SuSuSS$S:S:S~S~SgS$gSSS_S_SS$SRSRS[S4[SɧST ɧSq S q SMS MS%S %S_lS _lSYS YS)S)SS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations14o b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations14o b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations14o b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Y} u'C;_  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10vv  bb)N*N44>8?8IIS T ^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvbb N N  ::&&t t""$`%`'')L*L,,.8/8113$4$6689;;==?r@rBBD^E^GGIHJHLLN0O0QQSTVVXY[t[t]]_\`\bbdDeDggi,j,llnoqqssupvpxxzX{X}}@@((ll b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S;˃S;˃SS SSS:S:SSS=S=S+Sp+SSSSSRS`RS9\S9\SɧS$ ɧSq S q SGS GSS SYlS YlS)S )S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1YA  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1T} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1YA  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1T} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1YA  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1T k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 7[_?  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  ]])G*G44>1?1IIST^^hhrqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ss]] G G  11zzd d""$N%N'')8*8,,."/"113 4 6688:k;k==?U@UBBD?E?GGI)J)LLNOQQSSUrVrXXZ\[\]]_F`Fbbd0e0ggijllnoqsqsssuWvWxxz;{;}}uuYYˎː== b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S ˃S ˃SoS oSSS:S:S~S~SgSgSSpS_S_SSS RS` RSI[SI[SçS$ çSk S k SSS SS+S +SelS elSS SYSYSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Y}SwMqIm  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  bb)P*P44>>,,  ""$m%m'')[*[,,.I/I113747668%9%;;=>@@BCExExGGIdJdLLNNONQQS8T8VVX"Y"[[] ^ ``bbdkekggiUjUlln?o?qqs)t)vvxy{{}}rr\\чщFF00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}SA˃SA˃SS SSS:S:SSS=S=S1Sp1SSSSSURS`URS[S[SçS$ çSk S k SSS SS+S +SelS elSS SYSYSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1bH b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Y} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1bH b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Y} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1bH b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Y k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 642 0  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  gg)S*S44>=?=IIS%T%^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ssgg [ [  OOCC77++  "#%%' ( **,,.m/m113Y4Y668E9E;;=1>1@@BCEEG H J}J}LLNeOeQQSMTMVVX5Y5[[]^``bceyeyggiajallnIoIqqs1t1vvxy{{}~uu]]чщEE-- b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S˃S˃SiS iSSS}:S}:S~S~SgSgSSpSYSYSSSsRS`sRS[S[SçS$ çSk S k SAS ASS SSlS SlS#S #S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Y} q!9-Q  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  __)I*I44>1?1IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ss__ K K  77##q q""$\%\'')F*F,,.0/011346689;y;y==?c@cBBDMEMGGI5J5LLNOQQSTVtVtXXZZ[Z]]_@`@bbd&e&ggi j llnnpeqessuKvKxxz1{1}}ppVVɎɐ<< b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}SA˃SA˃SS SSS:S:SSS=S=S1Sp1SSSSSRS`RS'\S'\SçS$ çSk S k S;S ;SS SMlS MlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1bJ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Y} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1bJ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Y} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1bJ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Y k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 4XX||2  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  cc)K*K44>1?1IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1sscc S S  CC33##  "#%w%w'')_*_,,.G/G113/4/6689;;==?s@sBBD[E[GGIAJALLN%O%QQS T V{V{XXZ_[_]]_C`Cbbd'e'ggi j l}l}nnpaqassuEvExxz){)}} ccՉՋGG++ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS˃S˃SiS iSSS}:S}:S~S~SmSmSSpSYSYSSS=RS`=RSy[Sy[SS$ Se S e SGS GSS SYlS YlS_S _S/S/SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1k b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1k b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1k b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  0*$}  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  dd)P*P44>8?8IIS T ^^hiszsz} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1tthh \ \  PPDD44   " # %%'')n*n,,.Z/Z113F4F668292;;=>@@BCExExGGI`J`LLNHOHQQS0T0VVXY[[]^`t`tbbd\e\ggiDjDlln,o,qqstvvxxzp{p}}XX̂̄@@%% ~~ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSG˃SG˃SS SSS:S:SSSSS1Sp1SSSSSRS`RSC[SC[SS$ Se S e S/S /SS SAlS AlSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kQ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1=} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kQ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1=} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kQ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1= k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Y}SwMqGk  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  WW)A*A44>'?'IIS T ^^hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1llXX D D  00||f f""$P%P''):*:,,.$/$11346688:m;m==?S@SBBD9E9GGIJLLNOQxQxSSU^V^XXZD[D]]_*`*bbdeggiikilinnpOqOssu5v5xxz{}}ttXXʉʋ<<  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSʃSʃScS cSSSw:Sw:S~S~SmSmS梅Sp梅SSSSSSSRS`RS[S[SS$ Se S e S/S /SS SAlS AlS S  S߉S߉S_S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 >&J2V>b  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10rr  ^^)F*F44>*?*IIST^^hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1rrbb R R  BB22 z z""$b%b'')J*J,,.2/211346689;v;v==?Z@ZBBD>E>GGI"J"LLNOQxQxSSU\V\XXZ@[@]]_$`$bbdegzgziik^l^nnpBqBssu&v&xxz { }|}|``҄҆BB$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSG˃SG˃SS SSS:S:S#S#SSS7Sp7SSSSSRS`RSC[SC[SS$ Se S e SS SS S/lS /lS S  So߉So߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kS b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1=} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kS b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1=} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kS b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1= k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Y}SwMqGk  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  WW)=*=44>?IIST^r^rhhrRsR} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nnZZ F F  22zz` `""$F%F''),*,,,./11335k6k88:Q;Q==?3@3BBDEGGIIKhLhNNPJQJSSU,V,XXZ[]]__aabaddfCgCiik%l%nnpqsxsxuuwZxZzz|<}<nnދލNN b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃScS cSSSq:Sq:S~S~S[S[S驄Sp驄SMSMSSS7RS`7RSs[Ss[SS$ Se S e SMS MS%S %S_lS _lSS SwSwSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 +%~x  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  bb)T*T44>D?DIIS0T0^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ppbb T T  FF88**  "#%%'(*y*y,,.k/k113]4]668O9O;;=A>A@@B3C3EEG%H%JJLMOOQRTxTxVVXfYf[[]T^T``b@c@eeg*h*jjlmooqqsmtmvvxSyS{{}9~9uuYYˑ˓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSM˃SM˃SS SSS:S:S#S#S=S=S7Sp7SSSSSRS`RSC[SC[SS$ Se S e S5S 5S S  SGlS GlSAS ASSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1tZ& b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1b} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1tZ& b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1b} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1tZ& b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1b k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Y}SwMqGk  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  XX)B*B44>+?+IIST^^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kk[[ K K  ;;++ w w""$a%a'')K*K,,.5/51134668 9 ;~;~==?h@hBBDRERGGI;J;LLN#O#QQS T VVXXZg[g]]_O`Obbd7e7ggijllnoq{q{ssucvcxxzK{K}}33yycc b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃS]S ]SSSq:Sq:S~S~S[S[S驄Sp驄SMSMSSSESS`ESS\S\SS$ Se S e S5S 5S S  SGlS GlS5S 5SSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ae"Fb&  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  \\)J*J44>6?6IIST^^hispsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nn\\ J J  88&& y y""$g%g'')U*U,,.C/C1131416689;;= > @@BBDrErGGI^J^LLNHOHQQS2T2VVXY[[]^`{`{bbdaeaggiGjGlln+o+qqs t v~v~xxz`{`}}BB$$vvXX b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSM˃SM˃SS SSS:S:S)S)SCSCS=Sp=SSSSSaRS`aRS[S[SS$ Se S e S)S )SS S;lS ;lS S  S߉S߉SwS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t\( b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1c} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t\( b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1c} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t\( b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1c k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Y}m /C  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  VV)@*@44>(?(IIS T ^^hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kkWW C C  //{{e e""$O%O'')9*9,,.#/#113 4 6688:l;l==?V@VBBD@E@GGI(J(LLNOQQSSUgVgXXZM[M]]_3`3bbdeggiikmlmnnpQqQssu5v5xxz{}}ooSSƉƋ99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃS]S ]SSSk:Sk:S~S~SmSmS㩄Sp㩄SGSGSSS+RS`+RSg[Sg[SS$ S_ S _ SMS MS%S %S_lS _lSS SSS+S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 & ys  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  \\)L*L44>:?:IIS&T&^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ll\\ L L  <<,,   ""$t%t'')d*d,,.T/T113D4D668494;;=$>$@@BCEEGHJzJzLLNfOfQQSRTRVVX>Y>[[]*^*``bceeghjxjxllndodqqsPtPvvx} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}c/ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1>} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}c/ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1> k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Y}SwMqGk  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  UU)C*C44>1?1IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggUU C C  11  r r""$`%`'')N*N,,.@}@}BBDkEkGGIYJYLLNGOGQQS5T5VVX#Y#[[]^``bbdvevggidjdllnMoMqqs1t1vvxy{{}}kk݂݄OO33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃSWS WSSSk:Sk:S~S~SgSgS㩄Sp㩄SGSGSSSoSS`oSS\S\SS$ S_ S _ S5S 5S S  SGlS GlS;S ;S S SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Lp>b0T  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  VV)B*B44>,?,IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjVV B B  ..||h h""$T%T'')@*@,,.,/,11346689;z;z==?f@fBBDRERGGI} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}e1 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1>} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}e1 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1> k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Y}!Ei 1U  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  QQ);*;44>%?%IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggQQ ; ;  %%nnX X""$B%B''),*,,,./11346u6u88:_;_==?I@IBBD3E3GGIJLLNOQ|Q|SSUfVfXXZP[P]]_:`:bbd$e$ggijllnnpgqgssuKvKxxz/{/}}iiۉۋMM11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSQS QSSSe:Se:S~S~SUSUSݩSpݩSASASSS%RS`%RSa[Sa[SS$ SY S Y S;S ;SS SMlS MlSqS qSASASS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  ysm  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  VV)D*D44>0?0IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhVV D D  22  s s""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIXJXLLNBOBQQS,T,VVXY[[]^`u`ubbd_e_ggiIjIlln3o3qqstvvxy{|{|}}ffۂۄPPŇʼn::$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSS˃SS˃SS SSS:S:S/S/S1S1SCSpCSSSSSQS`QS7[S7[SS$ SY S Y S5S 5S S  SGlS GlSkS kS;S;SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1^} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1^} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l8 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1^ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Y}SwMqGk  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  [[)G*G44>1?1IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1gg[[ O O  CC77++  "#%%'')u*u,,.a/a113M4M668999;;=%>%@@BCEEGGIqJqLLNYOYQQSATAVVX)Y)[[]^``bbdmemggiUjUlln=o=qqs%t%vvx y {{}}ii݂݄QQŇʼn99!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSQS QSSS_:S_:S~S~SUSUSשSpשS;S;SSSuSS`uSS\S\SS$ SY S Y S#S #SS S5lS 5lSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1` b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1` b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1` b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 MqEi=a  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  SS)=*=44>%?%IIS T ^~^~hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++yye e""$P%P''):*:,,.$/$11346688:m;m==?W@WBBDAEAGGI)J)LLNOQQSSUhVhXXZN[N]]_4`4bbdeggijlslsnnpYqYssu?v?xxz%{%}} ~~dd׉׋JJ00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSY˃SY˃SS SSS:S:S/S/S7S7SCSpCSSSSSoSS`oSS\S\SS$ SY S Y SS SS S/lS /lS S  S߉S߉SkS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n: b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1_} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n: b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1_} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n: b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1_ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Y} /o!E7[  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  WW)?*?44>%?%IIS T ^{^{hhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggWW G G  77''  ""$k%k'')S*S,,.;/;113#4#668 9 ;;==?g@gBBDOEOGGI5J5LLNOQQSSUoVoXXZS[S]]_7`7bbdeggiikqlqnnpUqUssu9v9xxz{}}ssWWɉɋ;; b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃSKS KSSS_:S_:S~S~SaSaSשSpשS;S;SSSgRS`gRS[S[SS$ SY S Y S)S )SS S;lS ;lSGS GSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 g},B4X  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  XX)D*D44>,?,IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hh\\ P P  DD88((  "#%v%v'')b*b,,.N/N113:4:668&9&;;=>@@BBDlElGGITJTLLN?IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``LL 8 8  $$ppZ Z""$D%D'').*.,,./11346w6w88:a;a==?G@GBBD-E-GGIJLLNNPlQlSSURVRXXZ8[8]]_`bbdeggiikglgnnpMqMssu3v3xxz{}}rrVVȉȋ:: b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃSKS KSSSY:SY:S~S~SaSaSѩSpѩS5S5SSSQS`QS7[S7[SS$ SY S Y SS SS S)lS )lS S  S߉S߉SGS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Y}SwMqGk  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  RR):*:44>?IIST^x^xhhrZsZ} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&&n n""$V%V'')>*>,,.&/&11346688:j;j==?N@NBBD2E2GGIJLLNNPlQlSSUPVPXXZ4[4]]_`bbdegrgriikVlVnnp:q:ssuvxxz{}t}tXXʄʆ:: b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSY˃SY˃SS SSS:S:S5S5S S SISpISSSSS1RS`1RSm[Sm[SS$ SY S Y S S  SS SlS lS S  S]߉S]߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wC b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1=} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wC b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1=} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1wC b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1= k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 (&$"  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  KK)1*144>?IISS]n^nhhrNsN} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&nnT T""$:%:'') * ,,./1y1y335_6_88:E;E==?'@'BBD E GzGzIIK\L\NNP>Q>SSU V XXZ[]s]s__a]b]ddf?g?iik!l!nnpqststuuwVxVzz|8}8jjڋڍJJ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃSES ESSSY:SY:S~S~SOSOSѩSpѩS5S5SSSQS`QS+[S+[SS SS S| S SAS ASi~S i~SGNS GNSS SމSމS9S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1b n<`  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  VV)J*J44>:?:IIS&T&^^hiss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddVV H H  ::,,  "#%}%}'')o*o,,.a/a113S4S668E9E;;=7>7@@B)C)EEGHJJL M OOQQSnTnVVX\Y\[[]J^J``b6c6eeg h jjl m ooqqsotovvxUyU{{};~;!!ww[[͓͑ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LeS_˃S_˃SS SSS:S:S;S;S+S+SOSpOSSSSSURS`URS[S[SS SS S| S S/S /SW~S W~S5NS 5NS;S ;S1މS1މSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~J b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1\} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~J b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1\} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~J b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1\ k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1&Jj!#G  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  LL)@*@44>)?)IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__OO ? ?  //  k k""$_%_'')I*I,,.3/311346689;|;|==?f@fBBDPEPGGI9J9LLN!O!QQS T V}V}XXZe[e]]_M`MbbdAeAggi)j)llnoqqssumvmxxzU{U}}==%%mm b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LeSʃSʃS?S ?SSSS:SS:S~S~SOSOS˩Sp˩S/S/SSSQS`QSZSZSS SS S| S S)S )SQ~S Q~S/NS /NS)S )SމSމSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1Y}VzUyVz  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  PP)B*B44>.?.IIST^^histst} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$_%_'')M*M,,.;/;113)4)6689;;=>@|@|BBDjEjGGIVJVLLN@O@QQS*T*VVXY[[]]_s`sbbdYeYggi?j?lln/o/qqstvvxxzd{d}}FF(( zz\\ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LeS_˃S_˃SS SSS:S:S;S;S+S+SOSpOSSSSSQS`QS1[S1[SS SS S| S S#S #SK~S K~S)NS )NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1]} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1]} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1] k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations13 .0T  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  JJ)<*<44>$?$IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__KK 7 7  ##ooY Y""$K%K'')5*5,,./113 4 6~6~88:h;h==?R@RBBD4?4IIS"T"^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1^^LL : :  ((==+ +""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGI\J\LLNJOJQQS8T8VVX&Y&[[]^``bceyeyggigjglln\o\qqs@t@vvx$y${{}~zz^^ЇЉBB&&# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_Sw˃Sw˃SSdSSlS!S!S%S%S:S@:SMSMSSSS@SS SaSaSSSS SS S Sd  SUCS UCSoSS oSS\S \SSSmDžSmDžS)S)SuSuSs'S s'SASASSpSSlStSlSySTySSSwS$wSS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 [i6 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 [i6 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 [i6  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`&8\cTx>@d  "FoLp*QuDh b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@[Y}W{UySwQuOsMqKoI m  G k  E i  C g  A e % I  ?c `=a2V;_*o9]Ae7[7|5Y Nr3W1U/S-Q+O)M'K%I#G!ECA b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d[6lG"}X36{i @  D    J z U 0  fA bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  SS)E*E44>5?5IIS!T!^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33## { {""$/%/'')*,,./113=4=668-9-;;=>@@B C EEGGIuJuLLNaOaQQSMTMVVX9Y9[[]%^%``bceeggisjsllnkokqqsWtWvvxCyC{{}/~/}}iiߑߓ  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1_SʃSʃSS|SSSSS"S|"S1&SH1&S :S :So~SPo~SSXSSS+SP+SSSSSS$ SS, S S  SS$ S!S !SiDS| iDSgRSXgRS[S|[S뚅S뚅SoȅSxoȅS+S\+SSS'S'S#S8#SSS5lS5lSySySSlS߉S߉S}S } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1K bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1K  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1K bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`j;Bf9@d27>b&J5<`>b b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@`hpx ( 08@$ H  , P  4 X  < `  D h  LpTx\`d hlptx|i" g& e*c .a2_6_:] >   b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d`@ `@ `@  |  Z  8   m t HR#0~Yl bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10^^    ):*:44>&?&IIST^^hisrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1^^HH 2 2  {{&& ""$A%A'')+*+,,./11335t6t88:^;^==?H@HBBD2E2GGIJLLNOQ}Q}SSUgVgXXZQ[Q]]_;`;bbd%e%ggijllnoqtqtssuXvXxxz<{<}} vvZZ̎̐>># b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_Sw˃Sw˃SSdSSlS!S!S%S%S:S@:SSSSSSSS@SS SgSgSSSS SS S Sd  SUCS UCS1RS 1RSm[S m[SSSsDžSsDžS/S/SuSuSs'S s'S)S)SSpS;lSt;lSySTyS/S/S߉S$߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 [j6 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 [j6 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 [j6  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`"F`&*Nh&4Xr(>b|*% b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@[Y}W{UySwQuOsMqKoI m  G k  E i  C g  A e  ?c=a;_9]7[5Y3W D1U[/S-Q-Q#h-Q:^+O 0u)M'K%I#G!EC b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d[6lG"}X3i  D   z  U V 0    `f%A bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  MM)=*=44>)?)IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%ss_ _""$%''))+r,r..01346688:u;u==?a@aBBDMEMGGI9J9LLN!O!QQS T V}V}XXZe[e]]_M`Mbbd5e5ggijllnoqqssumvmxxzU{U}}==%% ii  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_SʃSʃSESdES Sl S!S!S#%S#%SS:S@S:S~S~SSSS@SS S˩S˩S/S/SS SS SSd S[CS [CSSS SS]S ]SMSMSsDžSsDžS/S/S{S{Sy'S y'S/S/SSpSAlStAlSySTySeSeS5S$5SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 \7 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 \7 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 \7  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`[*@du? >bh <`f [:^d? b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@[Y}W{UySwQuOsMqKoJ n  H l  F j  D h  B f  @d>b<`6Z:^0T8\*N6Z$H4XB2V<0T6.R 0,P*N(L&J$H"F DB b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d[6lG"}X3 i j  F C  !  | u W 2  hC bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  VV)B*B44>.?.IISS]^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbVV J J  >>22&&  "#%%'')p*p,,.\/\113H4H668494;;= > @@B C EEGGInJnLLNVOVQQS>T>VVX&Y&[[]^``bbd,e,ggijllnDoDqqs,t,vvxy{{}}ppXẊ̉@@((  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_Sq˃Sq˃SSdS Sl S!S!S#%S#%S:S@:SGSGSSSS@SS S[S[SSSS SS SSd S[CS [CSTS TSY]S Y]SSSyDžSyDžS5S5SuSuSs'S s'S/S/SSpSAlStAlSySTySeSeS5S$5SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 [m6 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 [m6 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 [m6  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`)>bh%IyNr7[ /4X~@fq/S b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@[Y}W{W{UySwQuOsMqK o  I m  G k  E i  C g  Ae@d?c D=a$p;_Rv9]2V7[65Yb3W1U/S-Q+O)M'K%I#G!E!E b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d[6lG"}X3 i X  D $   p z < U 0  fA bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  QQ)?*?44>-?-IIST^P^Phisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccQQ ? ?  --  n n""$\%\'')J*J,,.8/8113&4&6689;;=>@y@yBBDgEgGGIUJULLN?O?QQS)T)VVXY[[]]_r`rbbd\e\ggiFjFllnnpiqissuSvSxxz{{{}}eeڂڄOOćĉ99##  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_SʃSʃS?Sd?S Sl S!S!S#%S#%SS:S@S:S~S~SSSS@SS S˩S˩S/S/SS SS SSd S[CS [CSRS RS=[S =[SSSsDžSsDžS/S/SuSuSs'S s'SSSSpS)lSt)lSySTyS S S߉S$߉SeS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 \7 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 \7 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 \7  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`[<Dh6Y}>b 0W{8\*Uy2V b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@[Y}W{UySwQuOsMqKoI m  G k  E i  C g  A e  ?c=a;_9]7[5Y3W1U/S-Q+O*)M$}'Kw%Iq#Gm!EkCiA  b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d[6lG"}X3i  D   z U N 0 ' fYA2 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  RR):*:44>"?"IISS]^hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbRR B B  22"" z z""$f%f'')N*N,,.6/611346689;z;z==?b@bBBDJEJGGI2J2LLNOQQSSUlVlXXZP[P]]_4`4bgbgddfKgKiik/l/nnpZqZssu>v>xxz"{"}}xx\\Ή΋@@$$  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_Sq˃Sq˃SSdS Sl S!S!S#%S#%S:S@:SMSMSSSS@SS SaSaSSSS SS SSd S[CS [CSURS URS[S [SٚSٚSsDžSsDžS/S/SuSuSs'S s'SSSSpS)lSt)lSySTyS S S߉S$߉SwS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 [l6 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 [l6 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 [l6  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`%IbB 2Vo"F?c|&J&Ko*N2 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@[Y}W{UySwQuOsMqKoI m  G k  E i  C g  A e  ?c=a;_9]7[5Y3W1UBf/S"F-Q(t+OTx)M4X'K8%Id#G!ECA b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d[6lG"}X3i  D   z U  T 0  lf8A bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  NN)8*844>"?"IIST^@^@hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&tt` `""$K%K'')5*5,,./113 4 6~6~88:h;h==?R@RBBD?IIST^u^uhhr[s[} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__KK 7 7  ##ooY Y""$C%C'')-*-,,./11346v6v88:`;`==?H@HBBD.E.GGIJLLNNPmQmSSUSVSXXZ9[9]]_`bbdegxgxiik^l^nnpDqDssu*v*xxz{}}ii܄܆OO33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSe˃Se˃SSLSSTS!S!S)%S)%S:S(:SASASSS#S(#SSSUSUSSSS SS SSL SaCS aCSRS RSC[S C[SSSSSSS)SD)SSS;lS ;lSGSGSS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 [;: bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 [;: bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 [;:  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$\4XeZ~BBf,X|,P}X|:g b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@?5+!xn dZ~ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1zQ (  V  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  WW)C*C44>-?-IIST^^hhrqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1gg[[ O O  CC77''  ""$u%u'')a*a,,.M/M113949668%9%;;=>@@BBDmEmGGIUJULLN=O=QQS%T%VVX Y [[]]_i`ibbdQeQggi9j9lln!o!qqs t v}v}xxze{e}}MM55uu b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSESLESSTS!S!S)%S)%SY:S(Y:S~S~SSS#S(#SSSѩSѩS5S5SS SS SSL SaCS aCSSS SSK\S K\SSS/ƅS/ƅSSSSDSSSlS lS S SW߉S0W߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 [6 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 [6 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 [6  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`=Fj8\3<`)p2V 0(LvDh b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$H>4* wm cY} bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 - [2  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  JJ)0*044>?IISS]g^ghhrIsI} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%mmS S""$9%9'')*,,./1x1x335^6^88:D;D==?(@(BBD E G{G{IIK]L]NNP?Q?SSU!V!XXZ[]t]t__aVbVddf8g8iiklnnpprmsmuuwOxOzz|1}1eeՋՍEE b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSk˃Sk˃SSLSSTS!S!S/%S/%S:S(:SGSGSSS)S()SSS[S[SSS S  SS SSL SgCS gCSRS RS[S [S S SYƅSYƅSSSSDSSS#lS #lS S S߉S0߉SGS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 [;; bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 [;; bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 [;;  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`8kX|9b< Cl6Z)MvTx b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@PtFj<`2V(LB8 . bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 - [2  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  QQ)9*944>?IIST^u^uhhrYsY} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%%m m""$U%U'')=*=,,.%/%113 4 6688:i;i==?O@OBBD3E3GGIJLLNNPmQmSSUQVQXXZ5[5]]_`bbddfogoiikSlSnnp7q7ssuvxxzz|q}qUUDŽdž99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃSKSLKSSTS!S!S/%S/%S_:S(_:S~S~SSS)S()SSSשSשS;S;S S  SS SSL SgCS gCSyRS yRS[S [SSSSSSS)SD)SSS;lS ;lS/S/S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ]8 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ]8 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ]8  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`2Pt@)Vz(L /\4X5b Ae b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@(L> 0"wi[Mq?c bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1Y.  X -  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  OO)9*944>$?$IIS T ^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbRR B B  22""n n""$X%X'')B*B,,.,/,11346689;u;u==?_@_BBDIEIGGI4J4LLNOQQSTVxVxXXZ`[`]]_H`Hbbd0e0ggijllnoqtqtssu\v\xxzD{D}},,tt^^ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS_˃S_˃SSLS%ST%S!S!S;%S;%S:S(:S;S;SSS5S(5SSSOSOSSSS SS S+SL +SsCS sCSRS RS-\S -\SuSuSSSOSOSASDASSSSlS SlSSSkS0kSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{G b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 [bE bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{G b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 [bE bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{G  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 [bE  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$\+Op"=Pt>b -0T?~nB b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$ 0<$H0T<`HlTx bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1h J ,  p  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  YY)K*K44>=?=IIS)T)^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggYY K K  ==//!!  "#%~%~'')p*p,,.b/b113T4T668F9F;;=8>8@@B*C*EEGHJJL M OOQQSqTqVVX_Y_[[]M^M``b9c9eeg#h#jjl m ooqqsftfvvxLyL{{}2~2ppTTƑƓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃSKSLKSSTS!S!S/%S/%S_:S(_:S~S~SSS)S()SSSשSשS;S;S S  SS SSL SgCS gCS1RS 1RSm[S m[SSSŅSŅSSSSDSSS/lS /lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 \7 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 \7 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 \7  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`44X$&&J{"<}m   .{_ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@1U#G9+rdVz bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1  W ,V } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  MM)7*744>!?!IIST^x^xhhr^s^} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&rr\ \""$F%F'')0*0,,./11346y6y88:c;c==?M@MBBD7E7GGI!J!LLNOQzQzSSU`V`XXZF[F]]_,`,bbdeggiikflfnnpJqJssu.v.xxz{}}hhڄچNN44 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSe˃Se˃SSLS%ST%S!S!S;%S;%S:S(:SASASSS5S(5SSSUSUSSSS SS S+SL +SsCS sCSRS RSU[S U[SSSɅSɅSSS)SD)SSS;lS ;lS)S)S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 [aF bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 [aF bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 [aF  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`<[CgY}$HOsW{.RY}Uy8\c  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@` lx" .:"F. R :!^! bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1R 4xZ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>/?/IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIWJWLLNAOAQQS+T+VVXY[[]]_t`tbbdZeZggi@j@lln$o$qqstvwvwxxzY{Y}};;qqSS b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃSQSLQS1ST1S!S!SG%SG%Se:S(e:S~S~SSSAS(ASSSݩSݩSASAS!S !SS S7SL 7SCS CS]SS ]SS\S \SᛅSᛅSSSSS5SD5S S SGlS GlS_S_S/S0/SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  sN bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  sN bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  sN  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`+Bf|ImLpw Vz7[*` #G b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@`Z~TxNrHlBf<`6Z bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1f?  q J  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  PP)>*>44>.?.IIST^^hisrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$[%['')I*I,,.7/7113%4%6689;;=>@x@xBBDfEfGGIVJVLLNDODQQS2T2VVX Y [[]^``bbdsesggiajallnJoJqqs.t.vvxy{{}}hhڂڄLL00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSY˃SY˃SSLSSTS!S!S5%S5%S:S(:S5S5SSS/S(/SSSISISSSS SS S%SL %SmCS mCSRS RS \S  \SQSQS…S…SSSGSDGSSSYlS YlSSS}S0}S%S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p< b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 [;9 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p< b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 [;9 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p<  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 [;9  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`[&&J}Bu)M8k ] .a b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@`75YW{ y .,PNr bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1V C 0   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  WW)G*G44>7?7IIS#T#^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggWW G G  77''  ""$o%o'')_*_,,.O/O113?4?668/9/;;=>@@BCEEGGIwJwLLNcOcQQSOTOVVX;Y;[[]'^'``bceeggiujullnaoaqqsMtMvvx9y9{{}%~%ss__ՑՓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃSQSLQS1ST1S!S!SG%SG%Se:S(e:S~S~SSSAS(ASSSݩSݩSASAS!S !SS S7SL 7SCS CSyRS yRS[S [SSS_ƅS_ƅSSS#SD#SSS5lS 5lSSS߉S0߉S}S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  rM bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  rM bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  rM  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`))M|B;j*N)X||6ZFjjBf b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@0T*N$HB<6 0* bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 # | U. } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  LL)6*644>"?"IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbLL 6 6      iiS S""$=%='')'*',,./11335p6p88:Z;Z==?D@DBBD.E.GGIJLLNOQyQySSUcVcXXZM[M]]_7`7bbd!e!ggi j llnnpdqdssuHvHxxz,{,}}ff؉؋JJ.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS_˃S_˃SSLSSTS!S!S5%S5%S:S(:S5S5SSS/S(/SSSISISSSS SS S%SL %SmCS mCSRS RSO[S O[SSS˅S˅SSS/SD/SSSAlS AlS5S5SS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t@ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 [;: bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t@ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 [;: bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t@  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 [;:  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`[ //S[55Y[;;_[AAe b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@p%#GEig !>!!!<"`"""^###$ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 wdQ>+ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  QQ)=*=44>)?)IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))wwc c""$O%O'');*;,,.'/'11346688:u;u==?a@aBBDMEMGGI9J9LLN!O!QQS T V}V}XXZe[e]]_M`Mbbd5e5ggijllnoqyqyssuavaxxzI{I}}11uu]] b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃSWSLWSSTS!S!S5%S5%Sk:S(k:S~S~SSS/S(/SSS㩄S㩄SGSGSS SS S%SL %SmCS mCSSS SSQ\S Q\SSSSSISIS;SD;SSSMlS MlSqSqSAS0ASS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 \7 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 \7 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 \7  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$.Rw:^8\enBff,Lpp<` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@HlX|hx$4 D0T bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1T 8  d  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  ZZ)F*F44>2?2IIST^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffZZ N N  BB66**  "#%%'')t*t,,.`/`113L4L668898;;=$>$@@BCEEGGIrJrLLNZOZQQSBTBVVX*Y*[[]^``bbdnenggiVjVlln>o>qqs&t&vvxy{{}}jjނބRRƇƉ::"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSS˃SS˃SSLSSTS!S!S5%S5%S:S(:S/S/SSS/S(/SSSCSCSSSS SS S%SL %SmCS mCSmRS mRS[S [SSSąSąSSSASDASSSSlS SlSwSwSGS0GSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}e1 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 [[7 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}e1 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 [[7 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}e1  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 [[7  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`[&&J 2s4X#>@d;&JLp/S b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@\0Th(<`t4Hl bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 D < 4 , $ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  UU)C*C44>1?1IIST^^hiszsz} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggUU C C  11  r r""$`%`'')N*N,,.@}@}BBDkEkGGIYJYLLNCOCQQS-T-VVXY[[]^`v`vbbd`e`ggiJjJlln4o4qqstvvxy{}{}}}gg܂܄QQƇƉ;;%% b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃSWSLWSSTS!S!S5%S5%Se:S(e:S~S~SSS/S(/SSSݩSݩSASASS SS S%SL %SmCS mCSmRS mRS[S [SSSɅSɅSSS#SD#SSS5lS 5lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 \7 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 \7 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 \7  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H{=>q!E4g)M*]1U b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@dPt` p,<( L 8!\! bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1H ,tX } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  VV)>*>44>&?&IIS T ^|^|hhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&& ~ ~""$j%j'')R*R,,.:/:113"4"668 9 ;~;~==?f@fBBDNENGGI6J6LLNOQQSSUpVpXXZT[T]]_8`8bbdeggijlrlrnnpVqVssu:v:xxz{}}ttXXʉʋ<<  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSY˃SY˃SSLSSTS!S!S5%S5%S:S(:S/S/SSS/S(/SSSCSCSSSS SS S%SL %SmCS mCS+RS +RSg[S g[SSS΅S΅SaSaS)SD)SSS;lS ;lS S S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i5 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 [[7 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i5 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 [[7 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i5  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 [[7  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`[&&J_ ..Rc66Z i>>b b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@Tx (!L!!!`""" ###4$X$$$l%%&,&&& bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 | } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  RR)<*<44>&?&IIS T ^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **xxd d""$O%O'')9*9,,.#/#113 4 6688:l;l==?V@VBBD@E@GGI*J*LLNOQQSSUiViXXZO[O]]_5`5bbdeggijltltnnpZqZssu@v@xxz&{&}} ee؉؋KK11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSʃSʃS]SL]S%ST%S!S!S;%S;%Sq:S(q:S~S~SSS5S(5SSS驄S驄SMSMSS SS S+SL +SsCS sCSRS RS\S \SWSWS…S…SSS)SD)SSS;lS ;lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1` b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 \7 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1` b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 \7 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 \7  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`[&&J&.R)M&6Z](>b b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ DFjl#%IKoq(*N bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1B 1  ~  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  NN)8*844> ? IIST^y^yhhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''ss] ]""$G%G'')1*1,,./11346z6z88:d;d==?L@LBBD2E2GGIJLLNNPqQqSSUWVWXXZ=[=]]_#`#bbd e g|g|iikblbnnpHqHssu.v.xxz{}}mmSSʼnŋ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSM˃SM˃SSLSSTS!S!S5%S5%S:S(:S#S#SSS/S(/SSS7S7SSSS SS S%SL %SmCS mCSiSS iSS\S \S훅S훅SƅSƅS)S)S;SD;SSSMlS MlSYSYS)S0)SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1rZ& b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 [=6 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1rZ& b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 [=6 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1rZ&  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 [=6  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`%%IRv#GGkEii>bg@*N b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ Dn9c .X|#MqBf  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations12 5 8 ; > } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  [[)G*G44>1?1IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kk__ S S  GG;;++  "#%y%y'')e*e,,.Q/Q113=4=668)9);;=>@@BBDqEqGGIYJYLLNAOAQQS)T)VVXY[[]]_m`mbbdUeUggi=j=lln%o%qqs t vvxxzi{i}}QQłń99 yy b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSʃSʃS]SL]S%ST%S!S!S;%S;%Sk:S(k:S~S~SSS5S(5SSS㩄S㩄SGSGSS SS S+SL +SsCS sCSRS RSO[S O[SSS˅S˅SSSSDSSS)lS )lS S Si߉S0i߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1` b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 \7 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1` b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 \7 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 \7  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`[&&JY} ..RW{66ZUy>>b b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Ptv -/SUy { !2!!!4"X"""Z#~##$ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 m\K:) } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  NN)4*444>?IISS]k^khhrMsM} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))qqW W""$=%='')#*#,,. / 1|1|335b6b88:H;H==?,@,BBDEGGIIKaLaNNPCQCSSU%V%XXZ[]x]x__aZbZddf#?#IIST^y^yhhr]s]} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiYY I I  99))q q""$Y%Y'')A*A,,.)/)11346688:m;m==?S@SBBD7E7GGIJLLNNPqQqSSUUVUXXZ9[9]]_`bbdegsgsiikWlWnnp;q;ssuvxxz{}u}uYY˄ˆ== b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSʃSʃScSLcS%ST%S!S!S;%S;%Sq:S(q:S~S~SSS5S(5SSS驄S驄SMSMSS SS S+SL +SsCS sCSRS RS!\S !\SiSiSąSąSSS;SD;SSSMlS MlSGSGSS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 \7 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 \7 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 \7  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`[&&J8@d/S&JZ~ g8\t'% b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@BZ~4Xp2Jn $H bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 0 * $   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  SS)=*=44>(?(IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&&r r""$\%\'')F*F,,.0/011346689;y;y==?c@cBBDMEMGGI8J8LLN O QQSTV|V|XXZd[d]]_L`Lbbd4e4ggijllnoqxqxssu`v`xxzH{H}}00xxbb b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSG˃SG˃SSLS%ST%S!S!S;%S;%S:S(:SSSSS5S(5SSS1S1SSSS SS S+SL +SsCS sCS'SS 'SSc\S c\SSSȅSȅSSSSSDSS+S+SelS elSSS}S0}S%S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1gO b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 [[7 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1gO b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 [[7 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1gO  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 [[7  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$HCg8\\wp5 00TEiDhh b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@2VHl^4t&J< ` ! bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 . < JX } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  ]])O*O44>A?AIIS-T-^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kk]] O O  AA33%%  " # %%'')t*t,,.f/f113X4X668J9J;;=<><@@B.C.EEG H JJLMOOQQSuTuVVXcYc[[]Q^Q``b=c=eeg'h'jjlmooqqsjtjvvxPyP{{}6~6ttXXʑʓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSʃSʃS]SL]S%ST%S!S!S;%S;%Sq:S(q:S~S~SSS5S(5SSS驄S驄SMSMSS SS S+SL +SsCS sCSSS SS?\S ?\SSS΅S΅SgSgS/SD/SSSAlS AlS S S߉S0߉SwS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 \7 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 \7 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 \7  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`[&&J@Bf;_6Z\ {Ptv?' b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@`":^v !8!!!P"t""###*$N$$$f%%&(&&& bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 z } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  QQ);*;44>%?%IIS T ^|^|hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **vv` `""$J%J'')4*4,,./11346}6}88:g;g==?Q@QBBD;E;GGI%J%LLN O Q~Q~SSUdVdXXZJ[J]]_0`0bbdeggiikjljnnpNqNssu2v2xxz{}}llބކRRʼnŋ88 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSG˃SG˃SSLS%ST%S!S!S;%S;%S:S(:S#S#SSS5S(5SSS7S7SSSS SS S+SL +SsCS sCS?SS ?SS{\S {\SÛSÛSӅSӅS S S;SD;SSSMlS MlS;S;S S0 SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1iS b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 [[7 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1iS b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 [[7 bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1iS  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 [[7  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$HHl4XX| h"F&&J^6ZZ~  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@!!R"v"#(###h$$%>%%%~&&0'T''(((F)j))***\++,2, bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1ft } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  WW)E*E44>3?3IIST^^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiWW E E  33!!t t""$b%b'')P*P,,.>/>113,4,6689;;=>@@BBDmEmGGI[J[LLNEOEQQS/T/VVXY[[]^`x`xbbd^e^ggiDjDlln(o(qqs t v{v{xxz]{]}}??!!uuWW b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LS˃S˃S{S{SS$S:S:S~S~SmS$mSSSkSkSS$S+RS+RSg[S4g[SɧST ɧSq S q SYS YS1S 1SklS klSS SSSSSS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations14s b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations14s b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations14s b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 [[]_  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ww  ee)S*S44>A?AIIS/T/^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1wwee S S  AA//   ""$p%p'')^*^,,.L/L113:4:668(9(;;=>@@BCE{E{GGIiJiLLNWOWQQSETEVVX3Y3[[]!^!``bceeggitjtlln]o]qqsAtAvvx%y%{{} ~ {{__чщCC'' b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LSA˃SA˃SS SSS:S:SSS=S=S1Sp1SSSSS[RS`[RS[S[SɧS$ ɧSq S q SeS eS=S =SwlS wlSS SSSCS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[C b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1S} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[C b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1S} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[C b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1S k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 8642  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  dd)T*T44>B?BIIS.T.^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ttdd T T  DD44$$  "#%|%|'')l*l,,.\/\113L4L668<9<;;=,>,@@BCEEG H JJLLNnOnQQSZTZVVXFYF[[]2^2``bceeg h jjllnlolqqsXtXvvxDyD{{}0~0~~jj b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LS˃S˃SuSuSS$S:S:S~S~SmS$mSSSeSeSS$SRSRS[S4[SɧST ɧSq S q SGS GSS SYlS YlS)S )S߉S߉SS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations14q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations14q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations14q b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 [;?cg  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10uu  __)I*I44>3?3IIST^^hissss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1uu__ I I  33||f f""$P%P''):*:,,.$/$11346688:m;m==?W@WBBDAEAGGI+J+LLNOQQSSUtVtXXZ^[^]]_H`Hbbd2e2ggijllnoququssuYvYxxz={=}}!!ww[[͎͐?? b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LSA˃SA˃SS SSS:S:SSS=S=S1Sp1SSSSSRS`RS \S \SɧS$ ɧSq S q SMS MS%S %S_lS _lSSS SS#S#SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[E b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1R} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[E b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1R} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1[E b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1R k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 .RJnf  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  ``)L*L44>6?6IIST^^hiszsz} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1tt`` L L  88$$r r""$^%^'')J*J,,.6/6113"4"6689;;==?p@pBBD\E\GGIFJFLLN.O.QQSTVVXXZr[r]]_Z`ZbbdBeBggi*j*llnoqqssunvnxxzV{V}}>>&&jj b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S˃S˃SuS uSSS:S:S~S~SgSgSSpSeSeSSSRS`RS[[S[[SçS$ çSk S k SSS SS+S +SelS elSS S_S_SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 [Y}W{Uy  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ww  kk)W*W44>A?AIIS)T)^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1wwkk _ _  SSGG;;//  "###%%'(**,,.q/q113]4]668I9I;;=5>5@@B!C!EEG H JJLLNiOiQQSQTQVVX9Y9[[]!^!``b c e}e}ggiejellnMoMqqs5t5vvxy{{}~yyaaՇՉII11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}SG˃SG˃SS SSS:S:SSSCSCS1Sp1SSSSSURS`URS[S[SçS$ çSk S k SSS SS+S +SelS elSS SYSYSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1dL b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1[} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1dL b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1[} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1dL b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1[ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 3 1 / -  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  ^^)L*L44>8?8IIS"T"^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp^^ L L  ::(( { {""$i%i'')W*W,,.E/E113343668!9!;;=>@@BBDtEtGGI`J`LLNJOJQQS4T4VVXY[[]^`}`}bbdgegggiQjQlln;o;qqs%t%vvxy{{}}nnXX͇͉BB,, b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S ˃S ˃SoS oSSS:S:S~S~SgSgSSpS_S_SSSRS`RS[S[SçS$ çSk S k S;S ;SS SMlS MlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 [55YY}  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10uu  ee)M*M44>3?3IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1uuee U U  EE55%%  "#%y%y'')a*a,,.I/I1131416689;;=>@u@uBBD]E]GGICJCLLN'O'QQS T V}V}XXZa[a]]_E`Ebbd)e)ggi j llnnpcqcssuGvGxxz+{+}}ee׉׋II-- b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}SG˃SG˃SS SSS:S:S#S#SCSCS7Sp7SSSSSRS`RS[S[SçS$ çSk S k SAS ASS SSlS SlS#S #S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1dN b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1[} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1dN b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1[} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1dN b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1[ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 +OCg[ s  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  ]])G*G44>/?/IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qq]] I I  55!!  o o""$Z%Z'')D*D,,../.11346689;w;w==?a@aBBDKEKGGI3J3LLNOQQSSUrVrXXZX[X]]_>`>bbd$e$ggi j l}l}nnpcqcssuIvIxxz/{/}}nnTTǎǐ:: b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS ˃S ˃SoS oSSS}:S}:S~S~SmSmSSpSYSYSSSIRS`IRS[S[SçS$ çSk S k S5S 5S S  SGlS GlSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  -+)'  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  [[)E*E44>+?+IIST^^hhrhsh} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp\\ H H  44    j j""$T%T'')>*>,,.(/(11346688:q;q==?W@WBBD=E=GGI#J#LLN O Q|Q|SSUbVbXXZH[H]]_.`.bbdeggiikmlmnnpSqSssu9v9xxz{}}xx\\Ή΋@@$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSM˃SM˃SS SSS:S:S#S#SSS7Sp7SSSSSRS`RSU[SU[SçS$ çSk S k SAS ASS SSlS SlS_S _S/S/SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1mU! b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1;} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1mU! b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1;} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1mU! b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1; k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 [Y}W{Uy  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  ``)L*L44>4?4IIST^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ppdd X X  LL@@00  "#%~%~'')j*j,,.V/V113B4B668.9.;;=>@@BCEtEtGGI\J\LLNDODQQS,T,VVXY[[]]_p`pbbdXeXggi@j@lln(o(qqstvvxxzl{l}}TTȂȄ<<!!zz b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS˃S˃SiS iSSS}:S}:S~S~SmSmSSpSYSYSSSORS`ORS[S[SçS$ çSk S k S#S #SS S5lS 5lS S  Su߉Su߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1k b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1k b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1k b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 2 0 .,  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  YY)?*?44>!?!IIST^t^thhrTsT} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp\\ H H  44    ||b b""$H%H'').*.,,./11335m6m88:S;S==?5@5BBDEGGIIKjLjNNPLQLSSU.V.XXZ[]]__acbcddfEgEiik'l'nnp q szszuuw\x\zz|>}> ppPP b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSM˃SM˃SS SSS:S:S)S)SSS=Sp=SSSSSgRS`gRS[S[SçS$ çSk S k S/S /SS SAlS AlS S  S߉S߉S_S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1mW# b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1;} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1mW# b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1;} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1mW# b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1; k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 [ o 1!E  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  \\)D*D44>(?(IIS T ^~^~hhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp`` P P  @@00 x x""$`%`'')H*H,,.0/011346689;t;t==?X@XBBD[>]]_"`"bbdegxgxiik\l\nnp@q@ssu$v$xxz{}z}z^^ЄІ@@"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS˃S˃SiS iSSSw:Sw:S~S~SaSaS梅Sp梅SSSSSSSIRS`IRS[S[SS$ Se S e S;S ;SS SMlS MlSGS GSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 42 0 .  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10oo  \\)F*F44>/?/IIST^^hhrsss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1oo__ O O  ??// { {""$e%e'')O*O,,.9/9113#4#668 9 ;;==?l@lBBDVEVGGI?J?LLN'O'QQSTVVXXZk[k]]_S`Sbbd;e;ggi#j#lln o qqssugvgxxzO{O}}77}}gg b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSM˃SM˃SS SSS:S:S)S)SCSCS=Sp=SSSSSRS`RSO[SO[SS$ Se S e SMS MS%S %S_lS _lSS SwSwSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1v^* b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1d} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1v^* b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1d} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1v^* b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1d k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 [Y}W{Uy  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  ^^)P*P44>@?@IIS,T,^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ll^^ P P  BB44&&  " # %%'')u*u,,.g/g113Y4Y668K9K;;==>=@@B/C/EEG!H!JJLMOOQQStTtVVXbYb[[]P^P``b*?*IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mmYY E E  11}}g g""$Q%Q'');*;,,.%/%11346688:n;n==?X@XBBDBEBGGI*J*LLNOQQSSUiViXXZO[O]]_5`5bbdeggiikolonnpSqSssu7v7xxz{}}qqUUȉȋ;;!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSS˃SS˃SS SSS:S:S)S)SCSCS=Sp=SSSSSRS`RS9\S9\SS$ Se S e S5S 5S S  SGlS GlS5S 5SSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1v`, b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1e} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1v`, b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1e} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1v`, b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1e k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 [;_{?  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  ZZ)H*H44>4?4IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1llZZ H H  66$$ w w""$e%e'')S*S,,.A/A113/4/6689;;= > @@BBDpEpGGI\J\LLNFOFQQS0T0VVXY[[]^`y`ybbd_e_ggiEjElln)o)qqs t v|v|xxz^{^}}@@""ttVV b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃS]S ]SSSq:Sq:S~S~SmSmS驄Sp驄SMSMSSSCRS`CRS[S[SS$ S_ S _ S;S ;SS SMlS MlSkS kS;S;SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  / -+)  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  YY)G*G44>5?5IIS#T#^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kkYY G G  55##v v""$d%d'')R*R,,.@/@113.4.6689;;= > @@BBDoEoGGI]J]LLNKOKQQS9T9VVX'Y'[[]^``bcezezggihjhllnQoQqqs5t5vvxy{{}}ooSSŇʼn77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSS˃SS˃SS SSS:S:S/S/SSSCSpCSSSSSRS`RSO[SO[SS$ S_ S _ SGS GSS SYlS YlSS S}S}S%S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g3 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1<} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g3 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1<} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g3 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1< k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 [Y}W{Uy  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  XX)H*H44>6?6IIS"T"^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhXX H H  88((  ""$p%p'')`*`,,.P/P113@4@668090;;= > @@BCEEGHJvJvLLNbObQQSNTNVVX:Y:[[]&^&``bceeggitjtlln`o`qqsLtLvvx8y8{{}$~$rr^^ԑԓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃS]S ]SSSk:Sk:S~S~SmSmS㩄Sp㩄SGSGSSS{SS`{SS\S\SS$ S_ S _ S)S )SS S;lS ;lSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 OsGk?c  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  SS)=*=44>'?'IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiSS = =  ''ppZ Z""$D%D'').*.,,./11346w6w88:a;a==?K@KBBD5E5GGIJLLN O Q~Q~SSUhVhXXZR[R]]_<`*?*IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,zzf f""$R%R'')>*>,,.*/*11346689;x;x==?d@dBBDPEPGGI:J:LLN"O"QQS T V~V~XXZf[f]]_N`Nbbd6e6ggijllnoqzqzssubvbxxzJ{J}}22vv^^ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSWS WSSSk:Sk:S~S~SUSUS㩄Sp㩄SGSGSSS7RS`7RSs[Ss[SS$ S_ S _ S;S ;SS SMlS MlSqS qSASASS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 (&$"  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  __)K*K44>5?5IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kk__ S S  GG;;//##  "#%%'(*y*y,,.e/e113Q4Q668=9=;;=)>)@@BCEEGHJuJuLLN]O]QQSETEVVX-Y-[[]^``bbdqeqggiYjYllnAoAqqs)t)vvxy{{}}mmUUɇɉ==%% b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSY˃SY˃SS SSS:S:S5S5S7S7SISpISSSSS RS` RSI[SI[SS$ S_ S _ S5S 5S S  SGlS GlSkS kS;S;SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p< b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1`} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p< b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1`} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p< b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1` k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 [Y}W{Uy  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  RR)@*@44>,?,IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddRR @ @  ..  o o""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGITJTLLN>O>QQS(T(VVXY[[]]_q`qbbd[e[ggiEjElln/o/qqstvvxy{x{x}}bbׂׄLL66  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSWS WSSSe:Se:S~S~SUSUSݩSpݩSASASSSSS`SS\S\SS$ S_ S _ S#S #SS S5lS 5lS S  S߉S߉SqS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ^4t&J  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  YY)A*A44>'?'IIS T ^}^}hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiYY I I  99))   ""$m%m'')U*U,,.=/=113%4%668 9 ;;==?i@iBBDQEQGGI7J7LLNOQQSSUqVqXXZU[U]]_9`9bbdeggijlslsnnpWqWssu;v;xxz{}}uuYYˉˋ==!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSY˃SY˃SS SSS:S:S5S5S7S7SISpISSSSScSS`cSS\S\SS$ S_ S _ S#S #SS S5lS 5lSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1r> b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1a} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1r> b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1a} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1r> b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1a k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 [ /k?{+O  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  QQ);*;44>#?#IIS T ^|^|hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))wwc c""$N%N'')8*8,,."/"113 4 6688:k;k==?U@UBBD?E?GGI'J'LLN O QQSSUfVfXXZL[L]]_2`2bbdeggiikqlqnnpWqWssu=v=xxz#{#}} ||bbՉՋHH.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSQS QSSSe:Se:S~S~SaSaSݩSpݩSASASSSRS`RS[S[SS$ SY S Y SS SS S/lS /lS S  S߉S߉SeS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 r*.RVz~  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  OO)9*944>?IIST^z^zhhr^s^} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((tt^ ^""$H%H'')2*2,,./11346{6{88:e;e==?K@KBBD1E1GGIJLLNNPpQpSSUVVVXXZ<[<]]_"`"bbd e g}g}iikclcnnpIqIssu/v/xxz{}}nnRRĉċ66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS_˃S_˃SS SSS:S:S;S;S S SOSpOSSSSSRS`RS[S[SS$ SY S Y S/S /SS SAlS AlSMS MSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yE b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1;} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yE b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1;} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1yE b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1; k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 B<`Z~ x  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  TT)@*@44>(?(IIST^^histst} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddXX L L  @@44$$  ""$r%r'')^*^,,.J/J113646668"9";;= > @@BBDhEhGGIPJPLLN8O8QQS T VVXY[|[|]]_d`dbbdVeVggi>j>lln&o&qqstvvxxzj{j}}RRƂƄ::xx b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSKS KSSS_:S_:S~S~SaSaSשSpשS;S;SSSRS`RSC[SC[SS$ SY S Y S S  SS SlS lS S  S]߉S]߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 [Y}W{Uy  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  MM)3*344>?IISS]l^lhhrLsL} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((ppV V""$<%<'')"*",,./1{1{335a6a88:G;G==?)@)BBD E G|G|IIK^L^NNP@Q@SSU"V"XXZ[]u]u__a[b[ddf=g=iiklnnpqsrsruuwTxTzz|6}6hh؋؍HH b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS_˃S_˃SS SSS:S:S;S;S S SOSpOSSSSS=RS`=RSy[Sy[SS$ SY S Y SS SS S)lS )lS S  S߉S߉SMS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{G b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1;} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{G b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1;} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1{G b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1; k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  1 / -+  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP)8*844>?IIST^z^zhhr\s\} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$l l""$T%T'')<*<,,.$/$113 4 6688:h;h==?L@LBBD0E0GGIJLLNNPjQjSSUNVNXXZ2[2]]_`bbdegtgtiikXlXnnp r#  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  PP)<*<44>%?%IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33##o o""$[%['')E*E,,.///11346689;x;x==?b@bBBDLELGGI5J5LLNOQQSTVyVyXXZa[a]]_I`Ibbd=e=ggi%j%lln o qqssuivixxzQ{Q}}99!! ii b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LeSe˃Se˃SS SSS:S:S;S;S+S+SOSpOSSSSSRS`RSO[SO[SS SS S| S SGS GSo~S o~SMNS MNSS SމSމS?S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1^} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1^} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1N b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1^ k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations18DhX|  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  RR)N*N44>>?>IIS*T*^^hiss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``RR D D  66((   ""$%'')s*s,,.e/e113W4W668I9I;;=;>;@@B-C-EEGHJJL M OOQQSrTrVVX`Y`[[]N^N``b:c:eeg$h$jjlmooqrtstsvvxYyY{{}?~?%% {{__ёѓ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LeSʃSʃSES ESSSY:SY:S~S~SOSOSѩSpѩS5S5SSSQS`QSZSZSS SS S| S S#S #SK~S K~S)NS )NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1[\f  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  LL):*:44>"?"IIST^^hhrisi} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%qq[ [""$I%I'')3*3,,./11346|6|88:f;f==?P@PBBD:E:GGI"J"LLNOQ{Q{SSUaVaXXZG[G]]_-`-bbdeggijlslsnnpWqWssu;v;xxz{}}uuYỶ̋??%% b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LeSe˃Se˃SS SSS:S:SASAS+S+SUSpUSSSSSQS`QS[S[SS SS S| S S/S /SW~S W~S5NS 5NS/S /S%މS%މSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1_} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1_} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1_ k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1?><  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  NN)D*D44>0?0IIST^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``NN < <  **}}k k""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIXJXLLNBOBQQS,T,VVXY[[]^`u`ubbd[e[ggiAjAlln1o1qqstvvxxzf{f}}HH** ||^^ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations1_SʃSʃSS|SSS!S!S"S|"S7&SH7&S:S:Su~SPu~SSXSSS1SP1SSSSSS$ SS, SS SS$ S'S 'SoDS| oDS RSX RSI[S|I[SSSDžSxDžSS\SSS'S'SAS8ASSSSlSSlSySySSlSkSkSS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1M bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1M  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1M bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`]&&Jl _@dga8=a~ c7 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@*]u_Swa1Uc3~e\ g:^ i<km o   q   s   u   w  y{ }"$&(*, . 02468:<> @  b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d]Q:gt2Q. hE"  \  9   s P - gD bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  )N*N44>@?@IIS,T,^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``RR D D  66((SSE E""$%'')s*s,,.e/e113W4W668I9I;;=;>;@@B-C-EEGHJJLMOOQQStTtVVXbYb[[]P^P``b @  b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d]G: QtQ. jG$  ^  ;   u R / iF bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  YY)G*G44>3?3IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeYY M M  AA55))  "#%I%I'')5*5,,.!/!113M4M668999;;=%>%@@BCEEGGIsJsLLN[O[QQSCTCVVX+Y+[[]^``bbdoeoggiWjWllnKoKqqs3t3vvxy{{}~ww__ӇӉGG//  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1_SʃSʃSS|SSS!S!S"S|"S7&SH7&S :S :So~SPo~SSXSSS1SP1SSSSSS$ SS, SS SS$ S'S 'SoDS| oDSRSXRS[S|[SESES3˅Sx3˅SS\SSS(S(S)S8)SSS;lS;lSySyS)Sl)S߉S߉SS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1M bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1M  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1M bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`s*@dw .Bfl8\DhnfFjp! b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@k<6ZTxr%Ce=aCg[  ! E   # y n  ,   L p & J * N  D h  ,  x b3-QImc} .$H>bX| r#=3WMqg 2  (!L!!!B"f"""\## b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dkVA,}Hm^X)C.    i R ; $ v_H bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``    )D*D44>2?2IIST^^hiszsz} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``NN < <  **<<* *""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIZJZLLNDODQQS.T.VVXY[[]^`w`wbbd]e]ggiCjClln3o3qqstvvxxzh{h}}JJ,,bb# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_S}˃S}˃SSdS Sl S!S!S#%S#%S:S@:SYSYSSSS@SS SmSmSSSS SS SSd S[CS [CS7RS 7RSs[S s[SSSDžSDžSSSSS'S 'SSSSpS/lSt/lSySTyS S S߉S$߉SkS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ]f: bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ]f: bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ]f:  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` 11U|+3Z~~ -8\\ /::^ 1 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@]_ace g ikdm6Z  , o q   q C g    9 s ~   u P t   w  y{ }"$&(*, . 02468:<> @  b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d]:tQ[.  eh*Eo"  \  9   s P - gD bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  SS)?*?44>'?'IIS T ^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33## { {""$*%*'')*,,..013#4#668 9 ;;==?g@gBBDOEOGGI7J7LLNOQQSSUqVqXXZU[U]]_9`9bbdeggijlslsnnpcqcssuGvGxxz+{+}}ee׉׋II--  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_SʃSʃSKSdKSSlS!S!S)%S)%SY:S@Y:S~S~SSS#S@#SS SѩSѩS5S5SS SS SSd SaCS aCSSS SS\S \S S SDžSDžSSSSS'S 'S;S;SSpSMlStMlSySTySSSwS$wSS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ^; bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ^; bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ^;  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`]&&Ji77[^).Rl5Y"X| b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@)]r_Lpa&J$cmeGk g!E ihkm o   q   s   u   w  y{ }"$&(*, . 02468:<> @  b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d]N:`t)Q. hE"  \  9   s P - gD bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  PP)@*@44>0?0IISS]^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``PP @ @  00   x x""$h%h'')X*X,,.H/H113848668(9(;;=>@@BCEEGGIpJpLLN\O\QQSHTHVVX4Y4[[] ^ ``bbdBeBggi.j.llndodqqsPtPvvx###N$r$$ %%%&@& b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dtNh\jP8D8,   ~  t  j  `  VLB8.$ bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  MM)7*744>?IISS]z^zhhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&rr\ \""$F%F'')0*0,,./11346y6y88:c;c==?K@KBBD1E1GGIJLLNNPpQpSSUVVVXXZ<[<]]_"`"bWbWddf=g=iik#l#nnpIqIssu/v/xxz{}}nnTTƉƋ88  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_SʃSʃSESdESSlS!S!S)%S)%SY:S@Y:S~S~SSS#S@#SS SѩSѩS5S5SS SS SSd SaCS aCSRS RSU[S U[SSSDžSDžSSSSS'S 'S#S#SSpS5lSt5lSySTyS/S/S߉S$߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ^; bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1] b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ^; bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ^;  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`] /Jn~ _Pt&aVz,f\  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@]_ace g ikBfm@  o c   q = a    ; u    w ^   y 8 \  { }"$&(*, . 02468:<>@ B  b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d]:tQr.; hME"  \  9   s P - gD bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  LL)8*844>$?$IISS]^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``LL 8 8  $$rr^ ^""$J%J'')6*6,,."/"11346688:p;p==?\@\BBDHEHGGI4J4LLNOQQSTVxVxXXZ`[`]]_H`Hb{b{ddfcgciikKlKnoq|q|ssudvdxxzL{L}}44xx``  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_Sw˃Sw˃SSdSSlS!S!S)%S)%S:S@:SSSSSSS#S@#SS SgSgSSSS SS SSd SaCS aCSESS ESS\S \SɛSɛS7ʅS7ʅSSSSSSSQ(S Q(SSSSpSlStlSySTyS S SW߉S$W߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ]n: bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ]n: bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ]n:  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`,,Pu,P*Ossz)MMq!E''Ko: b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@g},B4XJn`v D%(v ;   Z ~ - Q > b  C g " F  Y }  * x    o \   4&J<`Nr` r 1C1UCgUygy&8&J8 \  J!n!!!\"" b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dgN5jTQ"8p> k P 5     dI.x bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  KK)1*144>?IISS]l^lhhrNsN} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&nnT T""$:%:'') * ,,./1y1y335_6_88:E;E==?)@)BBD E G|G|IIK^L^NNP@Q@SSU"V"XXZ[]u]u__abddffhmimkkmnpqsrsruuwTxTzz|6}6jjڋڍJJ  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSKSLKSSTS!S!S/%S/%S_:S(_:S~S~SSS)S()SSSשSשS;S;S S  SS SSL SgCS gCSSS SS\S \SSSSSSS/SD/SSSAlS AlSeSeS5S05SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ]: bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ]: bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ]:  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`]&&Jf;.R[)MZ~?cg b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@&mCg=Z~0T*qGk bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1IW } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  NN)<*<44>*?*IIST^^hhrsss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``NN < <  **}}k k""$Y%Y'')G*G,,.5/5113#4#6689;;==?v@vBBDdEdGGIRJRLLN bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ]9> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ]9>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`3]>b5_@d7aBfc<Gk b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$}wqk e_Y}Sw bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1Y2 d } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  RR)<*<44>'?'IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%%q q""$[%['')E*E,,.///11346689;x;x==?b@bBBDLELGGI7J7LLNOQQSTV{V{XXZc[c]]_K`Kbbd3e3ggijllnoqwqwssu_v_xxzG{G}}//wwaa b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSKSLKSSTS!S!S/%S/%S_:S(_:S~S~SSS)S()SSSשSשS;S;S S  SS SSL SgCS gCS1RS 1RSm[S m[SSSѴSѴSSSSDSSS)lS )lS S S߉S0߉SwS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ]: bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ]: bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ]:  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`]2Pt+eZ~5od ?yn" b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@A^4 X . u  K o ! E   bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1e,s: } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  MM)7*744>!?!IIST^z^zhhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%ss_ _""$J%J'')4*4,,./11346}6}88:g;g==?Q@QBBD;E;GGI%J%LLN O Q~Q~SSUdVdXXZJ[J]]_0`0bbdeggiikolonnpUqUssu;v;xxz!{!}}zz``ӉӋFF,, b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSq˃Sq˃SSLSSTS!S!S/%S/%S:S(:SMSMSSS)S()SSSaSaSSS S  SS SSL SgCS gCS1RS 1RSm[S m[SSSSS{S{SSDSSS)lS )lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ]9? bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ]9? bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ]9?  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$b$Hp (jPttr0TTx4z4X" b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@MqG k A e ; _ 5 Y / S )M#G bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1=oH! } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  NN)8*844>"?"IIST^y^yhhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''ss] ]""$G%G'')1*1,,./11346z6z88:d;d==?N@NBBD8E8GGI"J"LLNOQ{Q{SSUaVaXXZG[G]]_-`-bbdeggiikglgnnpKqKssu/v/xxz{}}iiۄۆOO‰‹55 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃSQSLQSSTS!S!S5%S5%Se:S(e:S~S~SSS/S(/SSSݩSݩSASASS SS S%SL %SmCS mCSORS ORS[S [SӚSӚSSSSS)SD)SSS;lS ;lSGSGSS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 _< bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 _< bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 _<  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$Hk4(Aet6;Tx84X: b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$t \Fj.R:"r Z~ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1P pB } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  TT)@*@44>*?*IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddXX L L  @@44$$  ""$r%r'')^*^,,.J/J113646668"9";;=>@@BBDjEjGGIRJRLLN:O:QQS"T"VVX Y [~[~]]_f`fbbdNeNggi6j6llnoqqstvzvzxxzb{b}}JJ22rr b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSe˃Se˃SSLS+ST+S!S!SA%SA%S:S(:SASASSS;S(;SSSUSUSSSS SS S1SL 1SyCS yCSSS SSW\S W\SSSӯSӯSmSmS5SD5S S SGlS GlS_S_S/S0/SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ]dI bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ]dI bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ]dI  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`]**N' 00TOs66Z<<`Ae  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$4 D0T@dPt` p bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1dH, } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>1?1IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIYJYLLNGOGQQS5T5VVX#Y#[[]^``bbdvevggidjdllnMoMqqs1t1vvxy{{}}kk݂݄OO33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃSQSLQSSTS!S!S/%S/%Se:S(e:S~S~SSS)S()SSSݩSݩSASAS S  SS SSL SgCS gCSURS URS[S [SٚSٚS뵅S뵅S3S3SSDSSS)lS )lS S S߉S0߉SGS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ^; bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ^; bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ^;  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`+<`~,Fjq4Pt{<$Z~ D b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@Bf*N 6  n  V z > b & J 2 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1b2R } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP)8*844>?IIST^t^thhrXsX} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$l l""$T%T'')<*<,,.$/$113 4 6688:h;h==?N@NBBD2E2GGIJLLNNPlQlSSUPVPXXZ4[4]]_`bbddfngniikRlRnnp6q6ssuvxxzz|p}pTTƄƆ88 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSk˃Sk˃SSLS+ST+S!S!SA%SA%S:S(:SASASSS;S(;SSSUSUSSSS SS S1SL 1SyCS yCSRS RSU[S U[SSSUSUSSS#SD#SSS5lS 5lSSS߉S0߉S}S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ]cJ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ]cJ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ]cJ  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$^$H{`>qb4g*d]  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ ,  < ( L 8 \ H l X|hx bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1tX<  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  MM)7*744>#?#IIS T ^^hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccMM 7 7  !!  jjT T""$>%>'')(*(,,./11335q6q88:[;[==?E@EBBD/E/GGIJLLNOQzQzSSUdVdXXZN[N]]_8`8bbd"e"ggi j llnnpeqessuIvIxxz-{-}}ggىًKK// b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃSWSLWS7ST7S!S!SM%SM%Sk:S(k:S~S~SSSGS(GSŔSŔS㩄S㩄SGSGS'S 'SS S=SL =SCS CSSS SS\S \SSSOSOSOSOSASDASSSSlS SlSSSkS0kSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  uR bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  uR bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  uR  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$HyTx:k",]RvOss Pt b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$"} {ywusq bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1[6l } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  VV)H*H44>:?:IIS&T&^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddVV H H  ::,,  "#%{%{'')m*m,,._/_113Q4Q668C9C;;=5>5@@B'C'EEGHJJL M OOQQSnTnVVX\Y\[[]J^J``b6c6eeg h jjl m o}o}qqsctcvvxIyI{{}/~/mmߌߎQQÑÓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkS_˃S_˃SSLSSTS!S!S5%S5%S:S(:S;S;SSS/S(/SSSOSOSSSS SS S%SL %SmCS mCSRS RS\S \S]S]S{S{SSS;SD;SSSMlS MlSqSqSAS0ASS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t@ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ]9= bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t@ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ]9= bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t@  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ]9=  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`]&&J46Z1U@Fji(LVz) b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$&JLpr)+OQuw . bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1o^M< } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  ]])I*I44>5?5IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ii]] Q Q  EE99--!!  "#%%'(*w*w,,.c/c113O4O668;9;;;='>'@@BCEEGGIuJuLLN]O]QQSETEVVX-Y-[[]^``bbdqeqggiYjYllnAoAqqs)t)vvxy{{}}mmUUɇɉ==%% b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃSWSLWS7ST7S!S!SM%SM%Se:S(e:S~S~SSSGS(GSŔSŔSݩSݩSASAS'S 'SS S=SL =SCS CSURS URS[S [SٚSٚSASASSS)SD)SSS;lS ;lS)S)S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1` b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  tQ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1` b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  tQ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  tQ  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$*Ny80Tk:6Z]<<``A b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@o m  k  i g e ca bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1G"}X3 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  RR)@*@44>.?.IIST^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddRR @ @  ..  o o""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGIVJVLLN@O@QQS*T*VVXY[[]]_s`sbbdYeYggi?j?lln#o#qqstvvvvxxzX{X}}::ppRR b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSe˃Se˃SSLS%ST%S!S!S;%S;%S:S(:S;S;SSS5S(5SSSOSOSSSS SS S+SL +SsCS sCS%RS %RSa[S a[SSSSSMSMS#SD#SSS5lS 5lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xD b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ]9> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xD b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ]9> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xD  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ]9>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`]&&Ja22Ve>>b i&JJn b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@0 T V z |  3 5 Y [8:^ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1+ xg } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  WW)?*?44>'?'IIS T ^}^}hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggWW G G  77''  ""$k%k'')S*S,,.;/;113#4#668 9 ;;==?g@gBBDOEOGGI7J7LLNOQQSSUqVqXXZU[U]]_9`9bbdeggijlslsnnpWqWssu;v;xxz{}}uuYYˉˋ==!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃS]SL]S%ST%S!S!S;%S;%Sq:S(q:S~S~SSS5S(5SSS驄S驄SMSMSS SS S+SL +SsCS sCSESS ESS\S \SɛSɛSSSSSGSDGSSSYlS YlSSS}S0}S%S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ^; bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ^; bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ^;  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$HEi((L ,,P%I 00Te) b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$8(L<`Ptdx&: bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1fL2 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  TT)D*D44>4?4IIS T ^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$ | |""$l%l'')\*\,,.L/L113<4<668,9,;;=>@@B C EEGGItJtLLN`O`QQSLTLVVX8Y8[[]$^$``bceeggirjrlln^o^qqsJtJvvx6y6{{}"~"pp\\ґғ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSY˃SY˃SSLS%ST%S!S!S;%S;%S:S(:S/S/SSS5S(5SSSCSCSSSS SS S+SL +SsCS sCSRS RS[S [S3S3S#S#SSS)SD)SSS;lS ;lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i5 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ]]; bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i5 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ]]; bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i5  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ]];  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H*N"FFjBfDhhZ~f r@ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$<`x:Rv,Ph*B f  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1ztnh } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  QQ);*;44>#?#IIS T ^|^|hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **vv` `""$J%J'')4*4,,./11346}6}88:g;g==?O@OBBD5E5GGIJLLNOQtQtSSUZVZXXZ@[@]]_&`&bbd e ggiikelennpKqKssu1v1xxz{}}ppVVȉȋ:: b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃS]SL]S%ST%S!S!S;%S;%Sk:S(k:S~S~SSS5S(5SSS㩄S㩄SGSGSS SS S+SL +SsCS sCSRS RS[S [S!S!SSSSS/SD/SSSAlS AlS;S;S S0 SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ^; bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ^; bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ^;  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H$H(0T8\,<`Lp 1Hl` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@* N > b R v f  z (<,P bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1~dJ0 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP)<*<44>(?(IIST^^hhrlsl} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((vvb b""$N%N''):*:,,.&/&11346688:t;t==?`@`BBDLELGGI8J8LLN O QQSTV|V|XXZd[d]]_L`Lbbd4e4ggijllnoqxqxssu`v`xxzH{H}}00tt\\ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSY˃SY˃SSLS%ST%S!S!S;%S;%S:S(:S5S5SSS5S(5SSSISISSSS SS S+SL +SsCS sCSURS URS[S [SٚSٚSSSSSSDSSS#lS #lS S Si߉S0i߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m9 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ]]; bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m9 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ]]; bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m9  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ]];  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H ."FFj 0Dhh2h4A b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@  ~  @ X |  2Vn 0Hl "F bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1b\VPJ  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  OO)5*544>?IISS]l^lhhrNsN} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **rrX X""$>%>'')$*$,,. / 1}1}335c6c88:I;I==?-@-BBDEGGIIKbLbNNPDQDSSU&V&XXZ[]y]y__a[b[ddf=g=iiklnnpqsrsruuwTxTzz|6}6jjڋڍJJ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSʃSʃScSLcS%ST%S!S!S;%S;%Sq:S(q:S~S~SSS5S(5SSS驄S驄SMSMSS SS S+SL +SsCS sCSRS RS\S \ScScSSSSSASDASSSSlS SlSwSwSGS0GSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ^; bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ^; bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ^;  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H0T::^Tx,PPt xBff . b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$*NTx ~7=ag &J bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1qbSD } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  RR)@*@44>.?.IIST^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddRR @ @  ..  o o""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGIVJVLLN@O@QQS*T*VVXY[[]]_s`sbbd]e]ggiGjGlln1o1qqstvvxy{z{z}}ddقلNNÇÉ88"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSS˃SS˃SSLS%ST%S!S!S;%S;%S:S(:S)S)SSS5S(5SSS=S=SSSS SS S+SL +SsCS sCS]SS ]SS\S \SᛅSᛅSѱSѱSSS;SD;SSSMlS MlSGSGSS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t^* b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ];: bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t^* b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ];: bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t^*  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ];:  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$HOs8\\p5Y 00TyDhh? b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$RvMqHlCg> b  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  VV)@*@44>+?+IIST^^hhrqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiYY I I  99)) u u""$_%_'')I*I,,.3/311346689;|;|==?f@fBBDPEPGGI;J;LLN#O#QQS T VVXXZg[g]]_O`Obbd7e7ggijllnoq{q{ssucvcxxzK{K}}33{{ee b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSʃSʃS]SL]S%ST%S!S!S;%S;%Sq:S(q:S~S~SSS5S(5SSS驄S驄SMSMSS SS S+SL +SsCS sCSRS RS[S [S3S3SSSeSeS/SD/SSSAlS AlSSS߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ^; bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ^; bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ^;  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H'K:Bf?c,P`W{Bf~ 1o b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@P t z  3 9 ] c "FLpx 1 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations15&y } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  QQ);*;44>%?%IIS T ^~^~hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))wwc c""$N%N'')8*8,,."/"113 4 6688:k;k==?U@UBBD?E?GGI)J)LLNOQQSSUhVhXXZN[N]]_4`4bbdeggijlslsnnpYqYssu?v?xxz%{%}} ~~dd׉׋JJ00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSS˃SS˃SSLS%ST%S!S!S;%S;%S:S(:S/S/SSS5S(5SSSCSCSSSS SS S+SL +SsCS sCS?SS ?SS{\S {\SÛSÛS]S]SSS/SD/SSSAlS AlS S S߉S0߉SwS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vb. b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ]:: bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vb. b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ]:: bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vb.  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ]::  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$HHl8\\ p"F 00T^Dhh  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ 9 ]  4 X /S*N|%Iw bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1#(-  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  RR)<*<44>&?&IIS T ^}^}hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++wwa a""$K%K'')5*5,,./113 4 6~6~88:h;h==?R@RBBD.?.IIST^^hhrrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hh\\ P P  DD88((  "#%v%v'')b*b,,.N/N113:4:668&9&;;=>@@BBDnEnGGIVJVLLN>O>QQS&T&VVXY[[]]_j`jbbdReRggi:j:lln"o"qqs t v~v~xxzf{f}}NN‚„66vv b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSG˃SG˃SSLS+ST+S!S!SA%SA%S:S(:S#S#SSS;S(;SSS7S7SSSS SS S1SL 1SyCS yCSSS SSQ\S Q\SSSySyS-S-SGSDGSSSYlS YlSqSqSAS0ASS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1iS b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ]]; bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1iS b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ]]; bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1iS  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ]];  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H@dPttp?,Eii8\4 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$h@8\Tx,p$ H  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 0@ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  WW)E*E44>5?5IIS#T#^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiWW E E  33!!t t""$b%b'')P*P,,.>/>113,4,6689;;=>@@BBDmEmGGI]J]LLNKOKQQS9T9VVX'Y'[[]^``bcezezggihjhllnQoQqqs5t5vvxy{{}}ooSSŇʼn77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSʃSʃScSLcS+ST+S!S!SA%SA%Sw:S(w:S~S~SSS;S(;SSS梅S梅SSSSSS SS S1SL 1SyCS yCSSS SSK\S K\SSSISISSS)SD)SSS;lS ;lS S S߉S0߉S_S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ^; bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ^; bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ^;  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`]&&J#*NVzCgRv  Bz)M2Nr b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ $ @ d  D `$@d D` bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1lhd`\  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  TT)<*<44>"?"IIST^x^xhhr\s\} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhXX H H  88((p p""$X%X'')@*@,,.(/(11346688:l;l==?R@RBBD6E6GGIJLLNNPpQpSSUTVTXXZ8[8]]_`bbdegrgriikVlVnnp:q:ssuvxxz{}t}tXXʄʆ<< b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSM˃SM˃SSLS+ST+S!S!SA%SA%S:S(:S)S)SSS;S(;SSS=S=SSSS SS S1SL 1SyCS yCS3SS 3SSo\S o\SSSSSSS5SD5S S SGlS GlSSS߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kW# b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ]]; bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kW# b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ]]; bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kW#  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ]];  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$HEiNrr=}=CggQu2 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@ d  \ 4x,PHl d< bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1P`p   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  QQ);*;44>'?'IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggQQ ; ;  %%nnX X""$B%B''),*,,,./11346u6u88:_;_==?I@IBBD3E3GGIJLLN O Q~Q~SSUhVhXXZR[R]]_<`M?MIIS9T9^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1yykk ] ]  OOAA33%%  "#%%' ( **,,.t/t113f4f668X9X;;=J>J@@B@?@IIS(T(^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvjj ^ ^  RRFF::..  ""#"%%'(**,,.p/p113\4\668H9H;;=4>4@@B C EEG H JJLLNhOhQQSPTPVVX8Y8[[] ^ ``bce|e|ggidjdllnLoLqqs4t4vvxy{{}~xx``ԇԉHH00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LS˃S˃S{S{SS$S:S:S~S~SmS$mSSSkSkSS$SRSRS\S4\SɧST ɧSq S q SMS MS%S %S_lS _lSMS MSSSS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations14s b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations14s b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations14s b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ]+O_  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ww  ee)S*S44>???IIS'T'^^h i sysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1wwee S S  AA//   ""$p%p'')^*^,,.L/L113:4:668(9(;;=>@@BCE{E{GGIgJgLLNQOQQQS;T;VVX%Y%[[]^``bbdjejggiPjPlln4o4qqstvvxxzi{i}}KK--aa b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LSG˃SG˃SS SSS:S:S#S#S=S=S7Sp7SSSSSRS`RS[S[SɧS$ ɧSq S q SAS ASS SSlS SlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]I b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1P} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]I b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1P} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1]I b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1P k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 %I5YEiUy  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10vv  ff)N*N44>4?4IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvff V V  FF66&&  "#%z%z'')b*b,,.J/J1132426689;;=>@v@vBBD^E^GGIDJDLLN(O(QQS T V~V~XXZb[b]]_F`Fbbd*e*ggijllnnpdqdssuHvHxxz,{,}}ff؉؋JJ.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S˃S˃S{S {SSS:S:S~S~SgSgSSpSeSeSSS1RS`1RSm[Sm[SçS$ çSk S k SeS eS=S =SwlS wlSS SSSCS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1r b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1r b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1r b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ]_ac  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10uu  ee)U*U44>C?CIIS/T/^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1uuee U U  EE55%%  "#%}%}'')m*m,,.]/]113M4M668=9=;;=->-@@BCEEG H JJLLNoOoQQS[T[VVXGYG[[]3^3``bceeg h jjllnmomqqsYtYvvxEyE{{}1~1 kk b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}SM˃SM˃SS SSS:S:S#S#SCSCS7Sp7SSSSS1RS`1RSm[Sm[SçS$ çSk S k S;S ;SS SMlS MlS S  S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1fP b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1]} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1fP b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1]} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1fP b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1] k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ]_ac  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10oo  ZZ)D*D44>*?*IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1oo[[ G G  33  i i""$S%S'')=*=,,.'/'11346688:p;p==?V@VBBD5?5IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ss__ K K  77##q q""$]%]'')I*I,,.5/5113!4!668 9 ;;==?o@oBBD[E[GGIEJELLN-O-QQSTVVXXZq[q]]_Y`YbbdAeAggi)j)llnoqqssumvmxxzU{U}}==%% ii b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}SM˃SM˃SS SSS:S:S)S)SCSCS=Sp=SSSSSORS`ORS[S[SçS$ çSk S k S)S )SS S;lS ;lS S  S{߉S{߉S#S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1fR b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1]} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1fR b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1]} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1fR b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1] k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ]iu+  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  ZZ)@*@44>"?"IIST^u^uhhrUsU} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qq]] I I  55!!  }}c c""$I%I'')/*/,,./11335n6n88:T;T==?6@6BBDEGGIIKkLkNNPMQMSSU/V/XXZ[]]__adbdddfFgFiik(l(nnp q s{s{uuw]x]zz|?}?!!qqQQ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS˃S˃SuS uSSS:S:S~S~SmSmSSpS_S_SSS+RS`+RSg[Sg[SçS$ çSk S k SMS MS%S %S_lS _lSS SYSYSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ]_ac  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  __)M*M44>9?9IIS#T#^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qq__ M M  ;;)) | |""$j%j'')X*X,,.F/F113444668"9";;=>@@BBDuEuGGIaJaLLNKOKQQS5T5VVXY[[] ^ `~`~bbdhehggiRjRlln.?.IIST^^hhrrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nn^^ N N  >>.. z z""$d%d'')N*N,,.8/8113"4"668 9 ;;==?k@kBBDUEUGGI>J>LLN&O&QQSTVVXXZj[j]]_R`Rbbd:e:ggi"j"lln o q~q~ssufvfxxzN{N}}66||ff b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS ˃S ˃SoS oSSS:S:S~S~SmSmSSpS_S_SSSRS`RS\S\SçS$ çSk S k S;S ;SS SMlS MlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ]+O_  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  \\)F*F44>.?.IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp\\ H H  44    n n""$Y%Y'')C*C,,.-/-11346689;v;v==?`@`BBDJEJGGI2J2LLNOQQSSUqVqXXZW[W]]_=`=bbd#e#ggi j l|l|nnpbqbssuHvHxxz.{.}}mmSSƎƐ99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSS˃SS˃SS SSS:S:S)S)SSS=Sp=SSSSSSS`SS?\S?\SçS$ çSk S k S5S 5S S  SGlS GlSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o[' b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations19} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o[' b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations19} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o[' b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations19 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ]!=a}A  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  YY)C*C44>+?+IIST^^hhrfsf} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nnZZ F F  22  ~~h h""$R%R'')<*<,,.&/&11346688:o;o==?Y@YBBDCECGGI+J+LLNOQQSSUjVjXXZP[P]]_6`6bbdeggiikplpnnpTqTssu8v8xxz{}}rrVVɉɋ<<"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS˃S˃SiS iSSS}:S}:S~S~SaSaSSpSYSYSSSURS`URS[S[SS$ Se S e SAS ASS SSlS SlSYS YS)S)SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  / 135  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  aa)M*M44>5?5IIST^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qqee Y Y  MMAA11  " # %%'')k*k,,.W/W113C4C668/9/;;=>@@BCEuEuGGI]J]LLNEOEQQS-T-VVXY[[]]_q`qbbdYeYggiAjAlln)o)qqstvvxxzm{m}}UUɂɄ==""{{ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSS˃SS˃SS SSS:S:S/S/SCSCSCSpCSSSSS%RS`%RSa[Sa[SS$ Se S e SAS ASS SSlS SlSkS kS;S;SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xb. b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1f} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xb. b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1f} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xb. b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1f k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ]_ac  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  XX)F*F44>4?4IIS"T"^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjXX F F  44""u u""$c%c'')Q*Q,,.?/?113-4-6689;;= > @@BBDnEnGGI\J\LLNJOJQQS8T8VVX&Y&[[]^``bceyeyggigjgllnPoPqqs4t4vvxy{{}}nnRRćĉ66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS˃S˃SiS iSSSw:Sw:S~S~SaSaS梅Sp梅SSSSSSSsRS`sRS[S[SS$ Se S e S)S )SS S;lS ;lS S  S߉S߉SYS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ?#G+O3W  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10oo  [[)C*C44>'?'IIS T ^}^}hhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1oo__ O O  ??// w w""$_%_'')G*G,,.///11346688:s;s==?W@WBBD;E;GGIJLLNOQuQuSSUYVYXXZ=[=]]_!`!bbdegwgwiik[l[nnp?q?ssu#v#xxz{}y}y]]τφ??!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSY˃SY˃SS SSS:S:S/S/SISISCSpCSSSSScSS`cSS\S\SS$ Se S e S/S /SS SAlS AlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xd0 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1g} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xd0 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1g} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xd0 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1g k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ] /g9qC  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  TT)>*>44>(?(IIST^^hhrhsh} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjTT > >  ((qq[ [""$E%E'')/*/,,./11346x6x88:b;b==?L@LBBD6E6GGI J LLN O QQSSUiViXXZS[S]]_=`=bbd'e'ggijllnnpjqjssuNvNxxz2{2}}llމދPPŽ44 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃScS cSSSw:Sw:S~S~SmSmS梅Sp梅SSSSSSSURS`URS[S[SS$ S_ S _ SGS GSS SYlS YlSS SwSwSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 79;=  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  __)Q*Q44>A?AIIS-T-^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mm__ Q Q  CC55''  " # %%'')v*v,,.h/h113Z4Z668L9L;;=>>>@@B0C0EEG"H"JJLMOOQQSuTuVVXcYc[[]Q^Q``b=c=eeg'h'jjlmooqqsjtjvvxPyP{{}6~6rrVVȑȓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSY˃SY˃SS SSS:S:S5S5SSSISpISSSSS%RS`%RSa[Sa[SS$ S_ S _ S;S ;SS SMlS MlSwS wSGSGSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1k7 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1:} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1k7 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1:} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1k7 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1: k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ]_ac  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  ^^)J*J44>4?4IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jj^^ R R  FF::..""  "#%%'(*x*x,,.d/d113P4P668<9<;;=(>(@@BCEEGHJtJtLLN\O\QQSDTDVVX,Y,[[]^``bbdpepggiXjXlln@o@qqs(t(vvxy{{}}llTTȇȉ<<$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃScS cSSSq:Sq:S~S~SmSmS驄Sp驄SMSMSSSSS`SSQ\SQ\SS$ S_ S _ S5S 5S S  SGlS GlS/S /S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 <`p .>b  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  YY)G*G44>3?3IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kkYY G G  55##v v""$d%d'')R*R,,.@/@113.4.6689;;= > @@BBDoEoGGI[J[LLNEOEQQS/T/VVXY[[]^`x`xbbd^e^ggiDjDlln(o(qqs t v{v{xxz]{]}}??!!ssUU b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSY˃SY˃SS SSS:S:S5S5SSSISpISSSSSiSS`iSS\S\SS$ S_ S _ S)S )SS S;lS ;lS S  S߉S߉SwS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m9 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1:} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m9 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1:} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m9 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1: k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ] 1mA}-Q  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  ZZ)B*B44>(?(IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjZZ J J  ::**   ""$n%n'')V*V,,.>/>113&4&6689;;==?j@jBBDRERGGI8J8LLNOQQSTVrVrXXZV[V]]_:`:bbdeggijltltnnpXqXssu>"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS]S ]SSSq:Sq:S~S~SUSUS驄Sp驄SMSMSSSORS`ORS[S[SS$ S_ S _ SGS GSS SYlS YlSS S}S}S%S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 3579  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ii  YY)I*I44>7?7IIS#T#^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1iiYY I I  99))   ""$q%q'')a*a,,.Q/Q113A4A668191;;=!>!@@BCEEGHJwJwLLNcOcQQSOTOVVX;Y;[[]'^'``bceeggiujullnaoaqqsMtMvvx9y9{{}%~%ss__ՑՓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS_˃S_˃SS SSS:S:S;S;S7S7SOSpOSSSSSRS`RS[S[SS$ S_ S _ SS SS S/lS /lS S  S߉S߉SeS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t@ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1b} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t@ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1b} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t@ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1b k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 @6ZPtj  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  NN)8*844>?IIST^w^whhr[s[} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''ss] ]""$G%G'')1*1,,./11346z6z88:d;d==?J@JBBD0E0GGIJLLNNPoQoSSUUVUXXZ;[;]]_!`!bbdegzgziik`l`nnpFqFssu,v,xxz{}}kkބކOO33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSWS WSSSk:Sk:S~S~SUSUS㩄Sp㩄SGSGSSSSS`SS]S]SS$ S_ S _ S/S /SS SAlS AlS5S 5SSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 c;3W   b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  SS)?*?44>)?)IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggSS ? ?  ++yye e""$Q%Q'')=*=,,.)/)11346689;w;w==?c@cBBDOEOGGI9J9LLN!O!QQS T V}V}XXZe[e]]_M`Mbbd5e5ggijllnoqyqyssuavaxxzI{I}}11uu]] b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkS_˃S_˃SS SSS:S:S;S;S=S=SOSpOSSSSSORS`ORS[S[SS$ S_ S _ S S  SS SlS lS S  S]߉S]߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vB b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1c} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vB b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1c} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vB b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1c k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  1357  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  NN)4*444>?IISS]i^ihhrIsI} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))qqW W""$=%='')#*#,,. / 1|1|335b6b88:H;H==?*@*BBD E G}G}IIK_L_NNPAQASSU#V#XXZ[]v]v__aXbXddf:g:iiklnnpprosouuwQxQzz|3}3eeՋՍEE b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSWS WSSSk:Sk:S~S~SaSaS㩄Sp㩄SGSGSSSCRS`CRS[S[SS$ SY S Y S5S 5S S  SGlS GlSkS kS;S;SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 , . 02  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>-?-IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIUJULLN?O?QQS)T)VVXY[[]]_r`rbbd\e\ggiFjFlln2o2qqstvvxy{{{{}}eeڂڄOOćĉ99## b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSe˃Se˃SS SSS:S:S;S;S S SOSpOSSSSSgRS`gRS[S[SS$ SY S Y S/S /SS SAlS AlS;S ;S S SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}I b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations19} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}I b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations19} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1}I b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations19 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  0<$H0T  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  OO)9*944>"?"IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbRR B B  22""n n""$X%X'')B*B,,.,/,11346689;u;u==?_@_BBDIEIGGI2J2LLNOQQSTVvVvXXZ^[^]]_F`Fbbd8e8ggi j llnoq|q|ssudvdxxzL{L}}44zzdd b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSQS QSSSe:Se:S~S~SaSaSݩSpݩSASASSSSS`SS;]S;]SS$ SY S Y S#S #SS S5lS 5lSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 l&JTx2  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP):*:44>"?"IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((vvb b""$M%M'')7*7,,.!/!113 4 6688:j;j==?T@TBBD>E>GGI&J&LLN O QQSSUeVeXXZK[K]]_1`1bbdeggiikplpnnpZqZssu@v@xxz&{&}} ee؉؋KK11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSe˃Se˃SS SSS:S:SASAS S SUSpUSSSSSURS`URS[S[SS$ SY S Y SS SS S/lS /lS S  S߉S߉SkS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations19} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations19} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1K b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations19 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 <>@B  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  MM)7*744>?IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&rr\ \""$F%F'')0*0,,./11346y6y88:c;c==?M@MBBD7E7GGIJLLNOQxQxSSU^V^XXZD[D]]_*`*bbdeggiiklllnnpPqPssu4v4xxz{}}nnRRʼnŋ88 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃSQS QSSS_:S_:S~S~SOSOSשSpשS;S;SSSsRS`sRS[S[SS SS S| S S5S 5S]~S ]~S;NS ;NSMS MSCމSCމSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1t 0<`6>b  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  UU)C*C44>+?+IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeYY M M  AA55%%  ""$u%u'')a*a,,.M/M113949668%9%;;=>@@BBDkEkGGISJSLLN;O;QQS#T#VVX Y [[]]_g`gbbd[e[ggiCjClln+o+qqstvvxxzo{o}}WW˂˄??$$ }} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LeSk˃Sk˃SS SSS:S:SASAS1S1SUSpUSSSSSQS`QS[S[SS SS S| S S;S ;Sc~S c~SANS ANSeS eS[މS[މSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1`} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1`} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1R b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1` k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1("F$H  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10^^  LL)D*D44>2?2IIS T ^^hiss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1^^LL : :  (({{i i""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIZJZLLNHOHQQS6T6VVX$Y$[[]^``bcewewggiejellnZoZqqs>t>vvx"y"{{}~xx\\·Ή@@$$ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LeSʃSʃSKS KSSS_:S_:S~S~SOSOSשSpשS;S;SSSQS`QSZSZSS SS S| S SS SE~S E~S#NS #NS S  S݉S݉SsS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1]w{ }  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  OO);*;44>?IIST^^hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33##k k""$W%W'')?*?,,.'/'11346688:k;k==?O@OBBD3E3GGIJLLNNPmQmSSUQVQXXZ5[5]]_`bbd e g{g{iik_l_nnpCqCssu'v'xxz { }}}}aaӄӆCC%% b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LeSk˃Sk˃SS SSS:S:SGSGS1S1S[Sp[SSSSSRS`RSU[SU[SS SS S| S S)S )SQ~S Q~S/NS /NS S  SމSމSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1a} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1a} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1T b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1a k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1%IImUy   b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10^^  HH):*:44>$?$IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1^^HH 2 2  {{eeO O""$A%A'')+*+,,./11335t6t88:^;^==?H@HBBD2E2GGIJLLNOQ{Q{SSUeVeXXZO[O]]_9`9bbd#e#ggi j llnoqrqrssuVvVxxz:{:}}ttXXʎʐ<< b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations1_SʃSʃSS|SSS!S!S"S|"S7&SH7&S:S:Su~SPu~SSXSSS1SP1SSSSSS$ SS, SS SS$ S'S 'SoDS| oDSRSXRS[[S|[[SSSuȅSxuȅS1S\1SSS'S'S/S8/SSSAlSAlSySyS;Sl;S S SS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1O bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1O  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1O bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`_68\ie>@d 4k6Z*q,P b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@_e kqw}$* 0 6   <   B  $ H  * N  0T Tx7[2V=a4Cg]Im;_Os=W{f]c iou{"( .4: @   b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d_>|[:xW6r =   t S  S   2   p O. l bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__    )@*@44>+?+IIST^^his}s}} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__OO ? ?  //??) )""$_%_'')I*I,,.3/311346689;|;|==?f@fBBDPEPGGI;J;LLN#O#QQS T VVXXZg[g]]_O`ObbdCeCggi+j+llnoqqssuovoxxzW{W}}??''qq# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_S}˃S}˃SSdSSlS!S!S)%S)%S:S@:SYSYSSS#S@#SS SmSmSSSS SS SSd SaCS aCSeTS eTS]S ]S霅S霅SoȅSoȅS+S+SSS'S 'S5S5S Sp SGlStGlSySTySkSkS;S$;SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 _a> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 _a> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 _a>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1` 00T:ePt!Lpp<2VVzp8\ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@_e kqw}$* 0 6   <   B  $ H  * N  a 0T3W6Z)n<`@dBf6{HlMqNrCTxZ~`f lrx~%+ 17 =   b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d_>|[:xWG6  Q  t   S \  2   p O. l bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  PP)@*@44>.?.IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$%'') * ,,..013'4'6689;;=>@z@zBBDhEhGGIVJVLLN@O@QQS*T*VVXY[[]]_s`sbbd]e]ggiGjGlln=o=qqs't'vvxy{{}}ppZZχωDD..  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1_SʃSʃSS|SSS!S!S"S|"S7&SH7&S:S:Su~SPu~SSXSSS1SP1SSSSSS$ SS, SS SS$ S'S 'SoDS| oDSSSXSSQ\S|Q\SSSͅSxͅSS\SSS})S})S#S8#SSS5lS5lSySyS Sl S߉S߉SkS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1O bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1O  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1O bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`~$$H D2;_o d ..Rt*Jn b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@v4Dhx6Fjz8Hl |  :  J n   ~  <  Lp >Nr @Lp|Dh8"FDh$otMq 0+O<` -x lVz(4Xd , P  \!!!"""$#H###T$x$$%%%&@& b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dvlbXND:0&    x i  l 4  ` JTH<0$ bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  )<*<44>&?&IIST^^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__KK 7 7  ##,, ""$K%K'')5*5,,./113 4 6~6~88:h;h==?R@RBBD bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 _b> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 _b>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$/Sd8 --Qv>+Ptt D*N&J b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@_e kqw}$* 0 6   <   B  $ H  * N  0T6Z<`BfHlNrTxZ~Z~,P"`gf9]  /ltrFj<x~%+ 17 =   b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d_>|[:xW6  t  S  2 !   f p + pP5/m bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  MM);*;44>%?%IIS T ^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%ss_ _""$ % ''))+j,j..013 4 6688:k;k==?U@UBBD?E?GGI)J)LLNOQQSSUhVhXXZN[N]]_4`4bbdeggijlslsnnpeqessuKvKxxz1{1}}ppVVɎɐ<<  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xeSʃSʃSKSdKSSlS!S!S/%S/%S_:S@_:S~S~SSS)S@)SS SשSשS;S;S S  SS SSd SgCS gCSRS RS-\S -\SuSuSuȅSuȅS1S1SSS'S 'S/S/SSpSAlStAlSySTySYSYS)S$)SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 `? bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 `? bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 `?  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`_7:^h#<?c?c .1U #GuC b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@_e kqw}$* 0 6   <   B  $ H  * N  0T4}6ZW{<`1UBf /x HlRvNr,PTx*Z~`f lrx~%+ 17 =   b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d_>|[:xW6]&  o  t 8  S   2   p O. l bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  MM);*;44>+?+IISS]^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__MM ; ;  ))||j j""$X%X'')F*F,,.4/4113"4"6689;;==?u@uBBDcEcGGISJSLLNAOAQQS/T/VVXY[[] ^ ``bbd.e.ggijllnQoQqqs5t5vvxy{{}}ooSSŇʼn77  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xeSw˃Sw˃SSdSSlS!S!S/%S/%S:S@:SSSSSSS)S@)SS SgSgSSS S  SS SSd SgCS gCS%RS %RSa[S a[SSS3˅S3˅SSSSS(S (S)S)SSpS;lSt;lSySTySASASS$SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 _q> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 _q> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 _q>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$`$Hrg .Vz|m:^asBEi b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@i4*NDh^x)C9 ]  S w   m     8  .RHl4Xb@*|z /b)MJnGk2Ve>60TNrl=7[Uys &   !D!!!>"b"""\## b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1diR;$ v_H1 4   T p $ [ t F 1 r]H bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  VV)B*B44>,?,IISS]^hhrrsr} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffZZ N N  BB66&&  ""$t%t'')`*`,,.L/L113848668$9$;;=>@@BBDlElGGITJTLLN|[:xW6  t  S  2 J    p \ O%.n l bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  II)3*344>?IISS]^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__II 3 3  ||ffP P""$:%:'')$*$,,./11335m6m88:W;W==?A@ABBD+E+GGIJLLNOQvQvSSU`V`XXZJ[J]]_4`4bfbfddfPgPiik:l:nnpiqissuMvMxxz1{1}}kk݉݋OO33  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xeS}˃S}˃SSdSSlS!S!S/%S/%S:S@:SYSYSSS)S@)SS SmSmSSS S  SS SSd SgCS gCSRS RS'\S '\SoSoSuȅSuȅS1S1SSS'S 'SSSSpS#lSt#lSySTyS S S߉S$߉SGS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 _p> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 _p> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 _p>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$x*Ni8,;_wLp  1Uy y 3W D b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@_e kqw}$* 0 6   <   B  $ H  * N  0T6Z<`BfHlNrTxZ~`'wf_ lGkr/S;x#~s% [+ 17 =   b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d_>|[:xW6  t  S  2   D p  dO4. l bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP)8*844>?IISS]x^xhhr\s\} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$l l""$T%T'')<*<,,.$/$113 4 6688:h;h==?N@NBBD2E2GGIJLLNNPlQlSSUPVPXXZ4[4]]_`bKbKddf/g/iiklnnp:q:ssuvxxz{}t}tXXʄʆ<<  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSQSLQSSTS!S!S/%S/%Se:S(e:S~S~SSS)S()SSSݩSݩSASAS S  SS SSL SgCS gCSSS SS\S \S/S/SSSSS/SD/SSSAlS AlSeSeS5S05SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 _> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 _> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 _>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`_ ..Rt===a++O~Gk=l%u b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@` 6Z 0wMq#Gd:^ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1F T  b  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  WW)C*C44>/?/IIST^^hhrsss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccWW K K  ??33''  "#%%'')q*q,,.]/]113I4I668595;;=!>!@@B C EEGGIoJoLLNWOWQQS?T?VVX'Y'[[]^``bbdkekggiSjSlln;o;qqs#t#vvx y {{}}ggۂۄOOÇÉ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSq˃Sq˃SSLSSTS!S!S5%S5%S:S(:SMSMSSS/S(/SSSaSaSSSS SS S%SL %SmCS mCSRS RS\S \SKSKSiSiSSS;SD;SSSMlS MlSSSeS0eS S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 _7B bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 _7B bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 _7B  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$`$H}Cw.R=rb7l  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ DB@><:86 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1" } X 3   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  XX)J*J44>7@@B)C)EEGHJJL M OOQQSpTpVVX^Y^[[]L^L``b8c8eeg"h"jjl m ooqqsetevvxKyK{{}1~1ooSSőœ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSQSLQSSTS!S!S/%S/%Se:S(e:S~S~SSS)S()SSSݩSݩSASAS S  SS SSL SgCS gCS=RS =RSy[S y[SSS-…S-…SSSSDSSS)lS )lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 _> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 _> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 _>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`_&&J  5l5YY}y DDh#( / b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@4{ Qu'K!h>b8Uy bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 ) p 7 ~E } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  SS);*;44>#?#IIST^y^yhhr]s]} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33## { {""$g%g'')O*O,,.7/711346689;{;{==?c@cBBDKEKGGI3J3LLNOQQSSUmVmXXZQ[Q]]_5`5bbdeggiikolonnpSqSssu7v7xxz{}}qqUUljNj99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSw˃Sw˃SSLSSTS!S!S5%S5%S:S(:SMSMSSS/S(/SSSaSaSSSS SS S%SL %SmCS mCS+RS +RSg[S g[SSS[DžS[DžS3S3S#SD#SSS5lS 5lS#S#S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 _7C bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 _7C bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 _7C  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$`$Ht f 0\ lDhhr,PPt b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@42 0 .,*(& bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 i DzU } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  RR)@*@44>.?.IIST^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddRR @ @  ..  o o""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGIVJVLLN@O@QQS*T*VVXY[[]]_s`sbbdYeYggi?j?lln#o#qqstvvvvxxzX{X}}::ppRR b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃSWSLWSSTS!S!S5%S5%Sk:S(k:S~S~SSS/S(/SSS㩄S㩄SGSGSS SS S%SL %SmCS mCSURS URS[S [SٚSٚSGSGSMSMSSDSSS/lS /lS S S߉S0߉SeS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 a@ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 a@ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 a@  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$Hv /4b5 Nrr;  0:^A b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@nZ~Fj2VB .l bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1"t F  j  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10``  KK)5*544>?IIST^v^vhhr\s\} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1``LL 8 8  $$ppZ Z""$D%D'').*.,,./11346w6w88:a;a==?I@IBBD/E/GGIJLLNNPnQnSSUTVTXXZ:[:]]_ ` bbdegygyiik_l_nnpEqEssu+v+xxz{}}jj݄݆PP‰‹44 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSk˃Sk˃SSLS1ST1S!S!SG%SG%S:S(:SGSGSSSAS(ASSS[S[SSS!S !SS S7SL 7SCS CS-SS -SSi\S i\SSSsSsSSSASDASSSSlS SlSSSwS0wSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 _fM bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 _fM bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 _fM  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`_&&J+66ZW{"FFj#2VVzOs b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ .B2VFjZ~n 0 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 v \ B (  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  VV)F*F44>6?6IIS"T"^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&& ~ ~""$n%n'')^*^,,.N/N113>4>668.9.;;=>@@BCEEGGIvJvLLNbObQQSNTNVVX:Y:[[]&^&``bceeggitjtlln`o`qqsLtLvvx8y8{{}$~$rr^^ԑԓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃSWSLWSSTS!S!S5%S5%Se:S(e:S~S~SSS/S(/SSSݩSݩSASASS SS S%SL %SmCS mCSRS RS[S [S3S3SąSąSgSgS SD SSSlS lS S S]߉S0]߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 `? bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 `? bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 `?  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H{B>q<`44XgZ~**N x b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@X|Dh0T@,~jVz bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1<  `2 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  KK)1*144>?IISS]h^hhhrJsJ} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&nnT T""$:%:'') * ,,./1y1y335_6_88:E;E==?)@)BBD E G|G|IIK^L^NNP@Q@SSU"V"XXZ[]u]u__aWbWddf9g9iiklnnpprnsnuuwPxPzz|2}2ff֋֍FF b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSq˃Sq˃SSLS1ST1S!S!SG%SG%S:S(:SGSGSSSAS(ASSS[S[SSS!S !SS S7SL 7SCS CSCRS CRS[S [SǚSǚSʅSʅSSS)SD)SSS;lS ;lS/S/S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 _eN bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 _eN bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 _eN  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`_&&J $k$H"w"F}!  D{ - b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ D4XHl\ p 2 "!F!!!6"Z" bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 tZ@& } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP)<*<44>(?(IIST^^hhrlsl} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((vvb b""$N%N''):*:,,.&/&11346688:t;t==?`@`BBDLELGGI8J8LLN O QQSTV|V|XXZd[d]]_L`Lbbd4e4ggijllnoqxqxssu`v`xxzH{H}}00tt\\ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃS]SL]S=ST=S!S!SS%SS%Sq:S(q:S~S~SSSMS(MS˔S˔S驄S驄SMSMS-S -SS SCSL CSCS CS'SS 'SSc\S c\SSSQSQSSS/SD/SSSAlS AlS;S;S S0 SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  wV bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  wV bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  wV  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$HAe""F} y D{9BBfyMq  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ace g ikmo bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 k H %   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  PP):*:44>%?%IIS T ^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccSS C C  33##o o""$Y%Y'')C*C,,.-/-11346689;v;v==?`@`BBDJEJGGI5J5LLNOQQSTVyVyXXZa[a]]_I`Ibbd1e1ggijllnoququssu]v]xxzE{E}}--uu__ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSe˃Se˃SSLS%ST%S!S!S;%S;%S:S(:S;S;SSS5S(5SSSOSOSSSS SS S+SL +SsCS sCSRS RS'\S '\SoSoS}ÅS}ÅS'S'S;SD;SSSMlS MlSqSqSAS0ASS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xD b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 _7A bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xD b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 _7A bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xD  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 _7A  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`_&&J!(LLp9]Nrr)ut7++O b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@u .4X^AGkq* bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1~ o ` Q B  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  TT)B*B44>0?0IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffTT B B  00  q q""$_%_'')M*M,,.;/;113)4)6689;;=>@|@|BBDjEjGGIXJXLLNBOBQQS,T,VVXY[[]^`u`ubbd_e_ggiIjIlln3o3qqstvvxy{|{|}}ffۂۄPPŇʼn::$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃS]SL]S=ST=S!S!SS%SS%Sk:S(k:S~S~SSSMS(MS˔S˔S㩄S㩄SGSGS-S -SS SCSL CSCS CS[RS [RS[S [SߚSߚSaDžSaDžS9S9S#SD#SSS5lS 5lS S S߉S0߉SkS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  vU bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  vU bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  vU  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H3,,P944XX|?<``!E b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@qsuwy{ }" bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 _ <vS } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  NN)8*844>"?"IIST^y^yhhr_s_} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccOO ; ;  ''ss] ]""$G%G'')1*1,,./11346z6z88:d;d==?N@NBBD8E8GGI"J"LLNOQ{Q{SSUaVaXXZG[G]]_-`-bbdeggiikglgnnpKqKssu/v/xxz{}}iiۄۆOO‰‹55 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSe˃Se˃SSLS%ST%S!S!S;%S;%S:S(:SASASSS5S(5SSSUSUSSSS SS S+SL +SsCS sCS7RS 7RSs[S s[SSS̅S̅SSS)SD)SSS;lS ;lS S S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|H b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 _7B bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|H b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 _7B bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|H  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 _7B  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`_&&Jg::^o*NNrw>bb b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@0TZ~=C g m!!"&""",#P###V$z$$% bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations13$w } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  QQ);*;44>%?%IIS T ^~^~hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))wwc c""$N%N'')8*8,,."/"113 4 6688:k;k==?U@UBBD?E?GGI)J)LLNOQQSSUhVhXXZN[N]]_4`4bbdeggijlslsnnpYqYssu?v?xxz%{%}} ~~dd׉׋JJ00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃScSLcS+ST+S!S!SA%SA%Sq:S(q:S~S~SSS;S(;SSS驄S驄SMSMSS SS S1SL 1SyCS yCSuSS uSS\S \SSS[S[SSS;SD;SSSMlS MlSeSeS5S05SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 `? bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 `? bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 `?  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$HPt44X!EEiBf1UUyy 4X b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@dX|p 8,PDh\ t bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1| d L 4   } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  QQ)?*?44>/?/IIST^^hissss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccQQ ? ?  --  n n""$\%\'')J*J,,.8/8113&4&6689;;=>@y@yBBDgEgGGIWJWLLNEOEQQS3T3VVX!Y![[]^``bbdtetggibjbllnKoKqqs/t/vvxy{{}}iiۂۄMM11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqS_˃S_˃SSLS+ST+S!S!SA%SA%S:S(:S5S5SSS;S(;SSSISISSSS SS S1SL 1SyCS yCSWSS WSS\S \SۛSۛSŅSŅSSS;SD;SSSMlS MlSSSSS#S0#SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m9 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 __? bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m9 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 __? bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m9  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 __?  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$HKo<``"x1U::^uRvv; b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@d D`$@d D`$ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 l h d ` \ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  ZZ)F*F44>0?0IIST^^histst} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jj^^ R R  FF::**  "#%x%x'')d*d,,.P/P113<4<668(9(;;=>@@BBDpEpGGIXJXLLN@O@QQS(T(VVXY[[]]_l`lbbdTeTggi>bOs4XX||  -or# b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$<0THl`x ( !@!!!4"X" bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 lT<$ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  MM)7*744>#?#IIS T ^^hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccMM 7 7  !!  jjT T""$>%>'')(*(,,./11335q6q88:[;[==?E@EBBD/E/GGIJLLNOQzQzSSUdVdXXZN[N]]_8`8bbd"e"ggi j llnnpeqessuIvIxxz-{-}}ggىًKK// b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqS_˃S_˃SSLS+ST+S!S!SA%SA%S:S(:S;S;SSS;S(;SSSOSOSSSS SS S1SL 1SyCS yCSsRS sRS[S [SSSυSυSSS#SD#SSS5lS 5lS S S߉S0߉SSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q= b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 __? bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q= b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 __? bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q=  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 __?  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H=,PQuC W{{%I4+O::^ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@@d D `!!"$"""@#d##$$$ %D%%%`&&'$''' bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1XTPLH } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  TT)<*<44>"?"IIST^x^xhhr\s\} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhXX H H  88((p p""$X%X'')@*@,,.(/(11346688:l;l==?R@RBBD6E6GGIJLLNNPpQpSSUTVTXXZ8[8]]_`bbdegrgriikVlVnnp:q:ssuvxxz{}t}tXXʄʆ<< b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwS˃S˃SiSLiS+ST+S!S!SA%SA%Sw:S(w:S~S~SSS;S(;SSS梅S梅SSSSSS SS S1SL 1SyCS yCSSS SSK\S K\SSSeÅSeÅS'S'SASDASSSSlS SlSwSwSGS0GSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 `? bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 `? bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 `?  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H<`"FFjlDhh(f4X b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@9]g",PZ~CMq{ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1l _ R E 8  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  [[)G*G44>3?3IIST^^hiswsw} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1gg[[ O O  CC77++  "#%%'')u*u,,.a/a113M4M668999;;=%>%@@BCEEGGIsJsLLN[O[QQSCTCVVX+Y+[[]^``bbdoeoggiWjWlln?o?qqs't'vvxy{{}}kk߂߄SSLJlj;;## b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSS˃SS˃SSLS+ST+S!S!SA%SA%S:S(:S/S/SSS;S(;SSSCSCSSSS SS S1SL 1SyCS yCSQSS QSS\S \S՛S՛SDžSDžSeSeSMSDMS%S%S_lS _lSSS}S0}S%S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vb. b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 _9> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vb. b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 _9> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1vb.  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 _9>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$HLpRvvC,PMqql> 0 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@KoLpMqNrOs bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1Z a h o v } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  \\)N*N44>@?@IIS,T,^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jj\\ N N  @@22$$  "#%%'')s*s,,.e/e113W4W668I9I;;=;>;@@B-C-EEGHJJLMOOQQStTtVVXbYb[[]P^P``b'?'IIS T ^}^}hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggWW G G  77''  ""$k%k'')S*S,,.;/;113#4#668 9 ;;==?g@gBBDOEOGGI7J7LLNOQQSSUqVqXXZU[U]]_9`9bbdeggijlslsnnpWqWssu;v;xxz{}}uuYYˉˋ==!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSY˃SY˃SSLS+ST+S!S!SA%SA%S:S(:S5S5SSS;S(;SSSISISSSS SS S1SL 1SyCS yCS3SS 3SSo\S o\SSS;҅S;҅SSS;SD;SSSMlS MlS5S5SS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xf2 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 _8> bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xf2 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 _8> bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xf2  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 _8>  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$HEi@dd(}= DDhQu` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@P t !!!Q"u"" ###R$v$$!%%%S&w&&"'''T(x((#)))U*y* bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1}  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  VV)D*D44>2?2IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhVV D D  22  s s""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIZJZLLNDODQQS.T.VVXY[[]^`w`wbbd]e]ggiCjClln'o'qqs t vzvzxxz\{\}}>> ttVV b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l}S˃S˃SiSLiS1ST1S!S!SG%SG%S}:S(}:S~S~SSSAS(ASSSSSYSYS!S !SS S7SL 7SCS CS SS  SSE\S E\SSSiŅSiŅSSS/SD/SSSAlS AlS S S߉S0߉SwS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 `? bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 `? bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 `?  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`_&&J#6ZZ~?cj((LC 8\ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$Dh *Nn4Tx:^~ bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 X V T R P } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  OO)9*944>!?!IIST^z^zhhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((tt^ ^""$H%H'')2*2,,./11346{6{88:e;e==?M@MBBD3E3GGIJLLNNPrQrSSUXVXXXZ>[>]]_$`$bbd e g}g}iikclcnnpIqIssu/v/xxz{}}nnTTƉƋ88 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l}SM˃SM˃SSLS1ST1S!S!SG%SG%S:S(:S)S)SSSAS(ASSS=S=SSS!S !SS S7SL 7SCS CSQSS QSS\S \S՛S՛SɅSɅSSSYSDYS1S1SklS klSSSS0S7S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kW# b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 __? bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kW# b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 __? bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1kW#  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 __?  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H=ahj@@d$,P8\\4 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$l"FDhf@> b !!! bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 H Z l ~ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  ZZ)J*J44>:?:IIS&T&^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjZZ J J  ::**   ""$r%r'')b*b,,.R/R113B4B668292;;=">"@@BCEEGHJzJzLLNfOfQQSRTRVVX>Y>[[]*^*``bceeghjxjxllndodqqsPtPvvx?IISS]l^lhhrNsN} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **rrX X""$>%>'')$*$,,. / 1}1}335c6c88:I;I==?-@-BBDEGGIIKbLbNNPDQDSSU&V&XXZ[]y]y__a[b[ddf=g=iiklnnpqsrsruuwTxTzz|6}6jjڋڍJJ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l}SS˃SS˃SSLS1ST1S!S!SG%SG%S:S(:S/S/SSSAS(ASSSCSCSSS!S !SS S7SL 7SCS CS'SS 'SSc\S c\SSSԅSԅSY SY SASDASSSSlS SlSGSGSS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m[' b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 __? bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m[' b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 __? bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m['  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 __?  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$HBfVzz%v%I4W{{Dh&&J b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@`""#:###$$8%\%%&&&Z'~'(4(((|))2*V** +++T,x, -.- bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1"4FXj } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  TT)@*@44>,?,IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,zzf f""$R%R'')>*>,,.*/*11346689;x;x==?d@dBBDPEPGGI8?8IIS T ^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1xxhh X X  HH88%%  ""$n%n'')X*X,,.B/B113,4,6689;;=>@u@uBBD_E_GGIHJHLLN0O0QQSTVVXY[t[t]]_\`\bbdDeDggi,j,llnoqqssupvpxxzX{X}}@@((pp b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LSM˃SM˃SS SSS:S:S#S#S=S=S7Sp7SSSSSsRS`sRS[S[SɧS$ ɧSq S q SYS YS1S 1SklS klSS S_S_SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_K b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1O} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_K b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1O} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_K b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1O k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 6<B$H  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  aa)O*O44>;?;IIS%T%^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ssaa O O  ==++ ~ ~""$l%l'')Z*Z,,.H/H113646668$9$;;=>@@BCEwEwGGIcJcLLNMOMQQS7T7VVX!Y![[] ^ ``bbdjejggiTjTlln>o>qqs(t(vvxy{{}}qq[[ЇЉEE// b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LS˃S˃SSSS$S:S:S~S~SmS$mS S SqSqSS$SSSSSK\S4K\SɧST ɧSq S q SGS GSS SYlS YlSS S߉S߉SS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations14u b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations14u b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations14u b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 _#?cC  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10vv  aa)K*K44>3?3IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvbb N N  ::&&p p""$Z%Z'')D*D,,../.11346689;w;w==?a@aBBDKEKGGI3J3LLNOQQSSUrVrXXZX[X]]_>`>bbd"e"ggijlxlxnnp\q\ssu@v@xxz${$}}zz^^щыDD** b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LSM˃SM˃SS SSS:S:S)S)S=S=S=Sp=SSSSSRS`RS[S[SɧS$ ɧSq S q SGS GSS SYlS YlS)S )S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_M b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1N} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_M b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1N} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1_M b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1N k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 A#G)M/S  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  ``)J*J44>2?2IIST^^hhrqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1tt`` L L  88$$r r""$]%]'')G*G,,.1/111346689;z;z==?d@dBBDNENGGI6J6LLNOQQSTVuVuXXZ[[[]]_A`Abbd'e'ggi j llnnpfqfssuLvLxxz2{2}}qqWWʎʐ== b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S˃S˃SS SSS:S:S~S~SgSgSSpSkSkSSS=RS`=RSy[Sy[SɧS$ ɧSq S q SSS SS+S +SelS elS}S }SMSMSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 _e kq  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  bb)P*P44>>?>IIS,T,^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ttbb P P  >>,,  ""$m%m'')[*[,,.I/I113747668%9%;;=>@@BCExExGGIfJfLLNTOTQQSBTBVVX0Y0[[]^``b c eeggiqjqllnZoZqqs>t>vvx"y"{{}~xx\\·Ή@@$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}SM˃SM˃SS SSS:S:S)S)SISIS=Sp=SSSSSRS`RS[S[SɧS$ ɧSq S q SMS MS%S %S_lS _lSeS eS5S5SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1hT  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1_} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1hT  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1_} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1hT  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1_ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 _99]]  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  cc)O*O44>7?7IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ssgg [ [  OOCC33  " # %%'')m*m,,.Y/Y113E4E668191;;=>@@BCEwEwGGI_J_LLNGOGQQS/T/VVXY[[]]_s`sbbd[e[ggiCjClln+o+qqstvvxxzo{o}}WW˂˄??$$ }} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S˃S˃S{S {SSS:S:S~S~SgSgSSpSkSkSSSRS`RS9\S9\SɧS$ ɧSq S q SAS ASS SSlS SlS#S #S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1r b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1r b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1r b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 _!9]u7  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10rr  \\)F*F44>0?0IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1rr\\ F F  00yyc c""$M%M'')7*7,,.!/!113 4 6688:j;j==?T@TBBD>E>GGI(J(LLNOQQSSUqVqXXZ[[[]]_E`Ebbd/e/ggijllnoqrqrssuVvVxxz:{:}}ttXXʎʐ<< b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}SS˃SS˃SS SSS:S:S)S)SISIS=Sp=SSSSSRS`RS-\S-\SɧS$ ɧSq S q S5S 5S S  SGlS GlS S  S߉S߉SkS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1hV" b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1_} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1hV" b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1_} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1hV" b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1_ k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 _3Wk+  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ss  __)G*G44>+?+IIST^^hhrcsc} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1sscc S S  CC33 { {""$c%c'')K*K,,.3/311346689;w;w==?[@[BBD?E?GGI#J#LLNOQyQySSU]V]XXZA[A]]_%`%bbd e g{g{iik_l_nnpCqCssu'v'xxz { }}}}aaӄӆCC%% b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S˃S˃SuS uSSS:S:S~S~SmSmSSpSeSeSSS=RS`=RSy[Sy[SçS$ çSk S k SSS SS+S +SelS elSS SYSYSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 _e kq  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  hh)T*T44>>?>IIS&T&^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1tthh \ \  PPDD88,,  " # %%' ( **,,.n/n113Z4Z668F9F;;=2>2@@BCEEG H J~J~LLNfOfQQSNTNVVX6Y6[[]^``bcezezggibjbllnJoJqqs2t2vvxy{{}~vv^^҇҉FF.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}SS˃SS˃SS SSS:S:S/S/SSSCSpCSSSSSCRS`CRS[S[SçS$ çSk S k SSS SS+S +SelS elSS S}S}S%S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q]) b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations17} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q]) b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations17} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q]) b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations17 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 _gow  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10oo  aa)S*S44>C?CIIS/T/^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ooaa S S  EE77))  " # %%'')x*x,,.j/j113\4\668N9N;;=@>@@@B2C2EEG$H$JJLMOOQRTwTwVVXeYe[[]S^S``b?c?eeg)h)jjlmooqqsltlvvxRyR{{}8~8ttXXʑʓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S˃S˃SuS uSSS:S:S~S~SmSmSSpS_S_SSS SS` SSE\SE\SçS$ çSk S k S;S ;SS SMlS MlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 _#?cC  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10rr  bb)J*J44>0?0IIST^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1rrbb R R  BB22""  "#%v%v'')^*^,,.F/F113.4.6689;;==?r@rBBDZEZGGI@J@LLN$O$QQSTVzVzXXZ^[^]]_B`Bbbd&e&ggi j l|l|nnp`q`ssuDvDxxz({(}} ~~bbԉԋFF** b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}SY˃SY˃SS SSS:S:S/S/SSSCSpCSSSSSiSS`iSS\S\SçS$ çSk S k SAS ASS SSlS SlSAS ASSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q_+ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations17} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q_+ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations17} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q_+ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations17 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 _ 1i;s!E  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10oo  ]])K*K44>7?7IIST^^hisqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1oo]] K K  99'' z z""$h%h'')V*V,,.D/D113242668 9 ;;=>@@BBDsEsGGI_J_LLNIOIQQS3T3VVXY[[]^`|`|bbdbebggiHjHlln,o,qqstvvxxza{a}}CC%%wwYY b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS ˃S ˃SoS oSSS:S:S~S~SaSaSSpS_S_SSSaRS`aRS[S[SS$ Se S e S/S /SS SAlS AlSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  -39?  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  XX)B*B44>(?(IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mmYY E E  11}}g g""$Q%Q'');*;,,.%/%11346688:n;n==?T@TBBD:E:GGI J LLNOQyQySSU_V_XXZE[E]]_+`+bbdeggiikjljnnpPqPssu6v6xxz{}}uuYYˉˋ==!! b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSY˃SY˃SS SSS:S:S5S5SISISISpISSSSS7RS`7RSs[Ss[SS$ Se S e SSS SS+S +SelS elSS SSS1S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zf2 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1h} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zf2 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1h} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zf2 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1h k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 _e kq  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  [[)K*K44>9?9IIS%T%^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kk[[ K K  ;;++   ""$s%s'')c*c,,.S/S113C4C668393;;=#>#@@BCEEGHJyJyLLNeOeQQSQTQVVX=Y=[[])^)``bceeghjwjwllncocqqsOtOvvx;y;{{}'~'uuaaבד b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS ˃S ˃SoS oSSS}:S}:S~S~SaSaSSpSYSYSSSmRS`mRS[S[SS$ Se S e SS SS S/lS /lS S  So߉So߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1l b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 9?!E'K  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  VV)<*<44>?IIST^q^qhhrQsQ} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mmYY E E  11yy_ _""$E%E'')+*+,,./11335j6j88:P;P==?2@2BBDEGGIIKgLgNNPIQISSU+V+XXZ [ ]~]~__a`b`ddfBgBiik$l$nnpqswswuuwYxYzz|;};mm݋ݍMM b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwSY˃SY˃SS SSS:S:S5S5SISISISpISSSSSWSS`WSS\S\SS$ Se S e S;S ;SS SMlS MlSAS ASSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zh4 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1i} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zh4 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1i} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zh4 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1i k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 _ /c3g7  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10kk  WW)C*C44>-?-IIST^^hhrqsq} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1kkWW C C  //}}i i""$U%U'')A*A,,.-/-11346689;{;{==?g@gBBDSESGGI=J=LLN%O%QQS T VVXXZi[i]]_Q`Qbbd9e9ggi!j!lln o q}q}ssuevexxzM{M}}55yyaa b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS˃S˃SiS iSSS}:S}:S~S~SmSmSSpSYSYSSSaRS`aRS[S[SS$ Se S e S5S 5S S  SGlS GlSAS ASSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1k b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1k b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1k b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 39?!E  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  YY)C*C44>,?,IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ll\\ L L  <<,, x x""$b%b'')L*L,,.6/6113 4 668 9 ;;==?i@iBBDSESGGI/?/IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggUU C C  11  r r""$`%`'')N*N,,.@}@}BBDkEkGGIWJWLLNAOAQQS+T+VVXY[[]]_t`tbbd^e^ggiHjHlln2o2qqstvvxy{{{{}}eeڂڄOOćĉ99## b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃScS cSSSw:Sw:S~S~SmSmS梅Sp梅SSSSSSSRS`RS[S[SS$ Se S e S)S )SS S;lS ;lS S  S߉S߉SqS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 )MAeY} q  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  UU)?*?44>'?'IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjVV B B  ..zzd d""$N%N'')8*8,,."/"113 4 6688:k;k==?U@UBBD?E?GGI'J'LLN O QQSSUfVfXXZL[L]]_2`2bbdeggiiklllnnpPqPssu4v4xxz{}}nnRRʼnŋ88 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS_˃S_˃SS SSS:S:S;S;SSSOSpOSSSSScSS`cSS\S\SS$ Se S e S)S )SS S;lS ;lSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q= b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations18} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q= b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations18} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q= b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations18 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 _ 1i;s!E  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  TT)>*>44>&?&IIS T ^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,zzf f""$Q%Q'');*;,,.%/%11346688:n;n==?X@XBBDBEBGGI*J*LLNOQQSSUiViXXZO[O]]_5`5bbdeggijltltnnpZqZssu@v@xxz&{&}} ee؉؋KK11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃScS cSSSw:Sw:S~S~S[S[S梅Sp梅SSSSSSSaRS`aRS[S[SS$ S_ S _ S;S ;SS SMlS MlSeS eS5S5SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 8> D&J  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  VV)D*D44>2?2IIS T ^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhVV D D  22  s s""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIZJZLLNHOHQQS6T6VVX$Y$[[]^``bcewewggiejellnNoNqqs2t2vvxy{{}}llނބPP‡‰44 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSe˃Se˃SS SSS:S:S;S;S=S=SOSpOSSSSSaRS`aRS[S[SS$ S_ S _ S/S /SS SAlS AlSMS MSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xD b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1d} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xD b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1d} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xD b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1d k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  .6>#G  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  WW)C*C44>+?+IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1gg[[ O O  CC77''  ""$u%u'')a*a,,.M/M113949668%9%;;=>@@BBDkEkGGISJSLLN;O;QQS#T#VVX Y [[]]_g`gbbdOeOggi7j7llnoqqstv{v{xxzc{c}}KK33qq b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃS]S ]SSSq:Sq:S~S~S[S[S驄Sp驄SMSMSSSoSS`oSS\S\SS$ S_ S _ S)S )SS S;lS ;lS S  S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 PtBf 4X  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  PP):*:44>$?$IIST^^hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffPP : :  $$mmW W""$A%A'')+*+,,./11335t6t88:^;^==?H@HBBD2E2GGIJLLNOQ{Q{SSUeVeXXZO[O]]_9`9bbd#e#ggi j llnnpfqfssuJvJxxz.{.}}hhډڋLL00 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSe˃Se˃SS SSS:S:SASAS=S=SUSpUSSSSSgRS`gRS[S[SS$ S_ S _ SS SS S)lS )lS S  S߉S߉SMS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zF b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1e} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zF b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1e} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zF b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1e k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 <B$H*N  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10gg  SS);*;44>?IIST^u^uhhrWsW} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ggWW G G  77''o o""$W%W'')?*?,,.'/'11346688:k;k==?O@OBBD3E3GGIJLLNNPmQmSSUQVQXXZ5[5]]_`bbddfogoiikSlSnnp7q7ssuvxxzz|q}qUUDŽdž77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS]S ]SSSk:Sk:S~S~SaSaS㩄Sp㩄SGSGSSS[RS`[RS[S[SS$ SY S Y S5S 5S S  SGlS GlSkS kS;S;SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 5;A#G  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  \\)H*H44>2?2IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hh\\ P P  DD88,,   "#%%'(*v*v,,.b/b113N4N668:9:;;=&>&@@BCEEGGIrJrLLNZOZQQSBTBVVX*Y*[[]^``bbdnenggiVjVlln@o@qqs(t(vvxy{{}}llTTȇȉ<<$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSk˃Sk˃SS SSS:S:SASAS S SUSpUSSSSSORS`ORS[S[SS$ SY S Y SAS ASS SSlS SlSS SkSkSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations17} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations17} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1M b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations17 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 * 06<  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  UU)G*G44>7?7IIS#T#^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccUU G G  99++  "#%z%z'')l*l,,.^/^113P4P668B9B;;=4>4@@B&C&EEGHJJLMO}O}QQSkTkVVXYYY[[]G^G``b3c3eeghjjlmooqqsjtjvvxPyP{{}6~6rrVVȑȓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃSWS WSSSk:Sk:S~S~SaSaS㩄Sp㩄SGSGSSSTS`TSM]SM]SS$ SY S Y S#S #SS S5lS 5lS S  S߉S߉SqS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 q-Q aA  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  VV)>*>44>$?$IIST^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&& ~ ~""$j%j'')R*R,,.:/:113"4"668 9 ;~;~==?f@fBBDNENGGI4J4LLNOQQSSUnVnXXZR[R]]_6`6bbdeggiikplpnnpXqXssu>"" b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSk˃Sk˃SS SSS:S:SGSGS S S[Sp[SSSSSRS`RS[S[SS$ SY S Y S)S )SS S;lS ;lS)S )S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations17} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations17} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations17 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 *N>bTxh  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10cc  QQ)?*?44>+?+IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ccQQ ? ?  --  n n""$\%\'')J*J,,.8/8113&4&6689;;=>@y@yBBDgEgGGISJSLLN=O=QQS'T'VVXY[[]]_p`pbbdVeVggi?IIST^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%ss] ]""$G%G'')1*1,,./11346z6z88:d;d==?J@JBBD0E0GGIJLLNNPoQoSSUUVUXXZ;[;]]_!`!bbdeggiiklllnnpRqRssu8v8xxz{}}ww[[͉͋??## b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LeSk˃Sk˃SS SSS:S:SGSGS1S1S[Sp[SSSSSQS`QS[S[SS SY S| Y SMS MSu~S u~SSNS SNSS SމSމSQS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1b} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1b} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1V b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1b k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1 /5;  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  OO)I*I44>7?7IIS#T#^^hiss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__OO ? ?  //w w""$q%q'')a*a,,.Q/Q113A4A668191;;=!>!@@BCEEGHJwJwLLNcOcQQSOTOVVX;Y;[[]'^'``bceeggiujullnmomqqsYtYvvxEyE{{}1~1 kk b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LeSʃSʃSQS QSSSe:Se:S~S~SOSOSݩSpݩSASASSSQS`QS[S[SS SY S| Y SS S9~S 9~SNS NS S  S݉S݉S1S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1`  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1_%+"F  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  NN)4*444>?IIST^u^uhhrUsU} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%qqW W""$=%='')#*#,,. / 1|1|335b6b88:H;H==?*@*BBD E G}G}IIK_L_NNPAQASSU#V#XXZ[]v]v__adbdddfFgFiik(l(nnp q s{s{uuw]x]zz|?}?!!qqQQ b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LeSq˃Sq˃SS SSS:S:SGSGS1S1S[Sp[SSSSSRS`RS \S \SS SY S| Y S5S 5S]~S ]~S;NS ;NS;S ;S1މS1މSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1c} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1c} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1X b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1c k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations10Tx8"  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10__  KK)?*?44>)?)IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1__KK 7 7  ##qq] ]""$Q%Q'')=*=,,.)/)11346689;w;w==?c@cBBDOEOGGI9J9LLN!O!QQS T V}V}XXZe[e]]_M`Mbbd5e5ggijllnoqqssumvmxxzU{U}}==%% ii b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations1_SʃSʃSS|SSS'S'S"S|"S=&SH=&S:S:S{~SP{~S!SX!SSS7SP7SSSSSS$ SS, SS SS$ S-S -SuDS| uDS1RSX1RSm[S|m[SSSɅSxɅSS\SSS'S'S5S85S S SGlSGlSySySMSlMSSSS } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Q bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Q  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1P  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Q bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`a -Fj mPt,wZ~ $HW{* b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$bm lGkv#G!j)Dh3B=e#G-Q7 [  A e  K o  U y  _  is}'3?'K3W?cKoW{c o{% 1=%I1 U  =!a! b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dbIC$[&fG( jK-   q  S  5  y[= bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa    )G*G44>1?1IIS%T%^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaUU I I  ==11SS? ?""$y%y'')e*e,,.Q/Q113=4=668)9);;=>@@BBDqEqGGIYJYLLNAOAQQS)T)VVXY[[]]_m`mbbdaeaggiIjIlln1o1qqstvvxy{u{u}}]]тфEE,,# b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_S˃S˃SSdSSlS!S!S/%S/%S:S@:S_S_SSS)S@)SS SsSsSSS S  SS SSd SgCS gCS1RS 1RSm[S m[SSSʅSʅSqSqS}S}S{(S {(SSSSpS/lSt/lSySTyS S S߉S$߉SkS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 a]B bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 a]B bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 a]B  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$b$Hk lAeev;;_7) b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$gk}Bf,<BY}4X/SJn)p`Fjv% ;  - Q  C g  Y }   o  4&J<`Vzp!;1UKoe 0&J@dZ~ t% ?  5!Y!!!O"s""# b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dgGN5VjQ8 o X A *   |eN7  bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa    )8*844> ? IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%// ""$G%G'')1*1,,./11346z6z88:d;d==?L@LBBD2E2GGIJLLNNPqQqSSUWVWXXZ=[=]]_#`#bbdeggiiknlnnnpTqTssu:v:xxz { }}yy__щыCC''  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1_SʃSʃSS|SSS'S'S"S|"S=&SH=&S:S:Su~SPu~S!SX!SSS7SP7SSSSSS$ SS, SS SS$ S-S -S MS|  MSSSTSS\S@\SS`SЅS8ЅSqSqS]S]S[*S4[*SSSSS/lS/lSyShyS S S߉S8߉SMS< } bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1Q bWuoperation.0ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue4bytes2560patternContiguousinvocations1 bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1Q  bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1  bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1O  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1Q bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b)goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1D33WZ~!!E+Y}a2h b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b)goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1$&Jp9_(Nr?c=a=   ` , P : ^  v  ?  e  .TxCg 2V|Ce,Nr7[} Df -Os  8!\!!!~""!#E###g$$ %.%%%P&t&&'''9(](() b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1do8J         bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  )=*=44>#?#IIST^^hhrisi} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaQQ A A  11!!<<$ $""$Y%Y'')A*A,,.)/)11346688:m;m==?S@SBBD7E7GGIJLLNNPqQqSSUUVUXXZ9[9]]_`bbd e ggiikclcnnpGqGssu+v+xxz{}}eeׄ׆II++ # b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 $ bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1x_S˃S˃SSdSSlS!S!S/%S/%S;S@;SeSeSSS)S@)SS SySySSS S  SS SSd SgCS gCSSS SS\S \S;S;S҅S҅SSS S S+S +S S SSpSlStlSySTyS S S]߉S$]߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1x b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 a^B bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1x b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 a^B bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1x  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 a^B  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`2LpHl%Rv"+X|j 1^ D b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@8\ BfLpVz@(]` 3 W  2 -  t  j J n  < D  t  "F~,P6Z@dHlLpPt Tx$X|(\ !,!!!`"" #0###d$$%4%%%h&&'8'''l(()<)))p**+@+ b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1d (d2+<r9FPZ  d  l  t  |  $,4 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa    )4*444>?IIST^w^whhrYsY} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%,, ""$=%='')#*#,,. / 1|1|335b6b88:H;H==?,@,BBDEGGIIKaLaNNPCQCSSU%V%XXZ[]x]x__afbfddfHgHiik*l*nnp q s}s}uuw_x_zz|A}A##uuUUŐŒ  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xeSʃSʃSQSdQSSlS!S!S5%S5%Se:S@e:S~S~SSS/S@/SS SݩSݩSASASS SS S%Sd %SmCS mCSURS URS[S [SٚSٚSȅSȅSSSSS'S 'S;S;SSpSMlStMlSySTySSSkS$kSS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 bC bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 bC bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 bC  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`a 0Fj{uPtz##Z~7BHl'K b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$bl lDhv@`)8\34|= Tx#G-Q7 [  A e  K o  U y  _  is}& 0: D*N4X>bHlRv\fpz# - 7  !A! b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dbHC$X fG( jK,   n  O  0  rS4 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  SS)E*E44>7?7IISS]^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaSS E E  77))   ""$x%x'')j*j,,.\/\113N4N668@9@;;=2>2@@B$C$EEGHJJLMO}O}QQSkTkVVXYYY[[]G^G``bbddedggiNjNllnoqqsjtjvvxPyP{{}6~6ttXXʑʓ  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xeS}˃S}˃SSdSSlS!S!S5%S5%S:S@:SYSYSSS/S@/SS SmSmSSSS SS S%Sd %SmCS mCS7RS 7RSs[S s[SSSȅSȅSSSSS'S 'S#S#SSpS5lSt5lSySTyS/S/S߉S$߉SS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 asB bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 asB bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1i  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 asB  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$b$Hv l4b vNrr :^^) b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$bv lbvNr:^)&J36="t#G-Q7 [  A e  K o  U y  _  is}& 0: D*N4X>bHlRv\fpz# - 7  !A! b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1dbRC$$vHfG( jK,   n  O  0  rS4 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  QQ);*;44>&?&IISS]^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddTT D D  44$$p p""$Z%Z'')D*D,,../.11346689;w;w==?a@aBBDKEKGGI6J6LLNOQQSTVzVzXXZb[b]]_J`Jb|b|ddfdgdiikLlLnoqxqxssu`v`xxzH{H}}00xxbb  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xeSʃSʃSQSdQSSlS!S!S5%S5%Se:S@e:S~S~SSS/S@/SS SݩSݩSASASS SS S%Sd %SLS LSsRS sRS[S [SSSȅSȅSSXSS|S'S'S)S4)SSS;lS;lSySyS#Sh#S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 bC bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 bC bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1a   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 bC  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b)goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1D*+O68"Fa2VBf b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b)goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1$a ku(2<"F,P,P(p6 Z  H l  @ d D  J n  T x  ^  hr|% /9C)M3W=aGkQu[eoy", 6  !@! b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1daB#ehF0'x@jK,   n  O  0  rS4 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  OO)=*=44>+?+IISS]^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaOO = =  ++~~l l""$Z%Z'')H*H,,.6/6113$4$6689;;=>@w@wBBDeEeGGISJSLLN=O=QQS'T'VVXY[[]]_p`pbbdeggiikjljn(o(qqs t v{v{xxz]{]}}??!!uuWW  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1xeS˃S˃SSdSSlS!S!S5%S5%S:S@:SYSYSSS/S@/SS SmSmSSSS SS S%Sd %SmCS mCSRS RS[S [SSSȅSȅSSSSS'S 'SSSSpS)lSt)lSySTyS S S߉S$߉SeS(} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 arB bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 arB bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 arB  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`)m6Z"8\v D:^lBf <`ed b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@a ku(2<"F`,PLp6 Z 8 \  @ d $ H  J n  4  T x r  ^ ^   hJnr|% /9C)M3W=aGkQu[eoy", 6  !@! b-g operation.4ipu_stack_rearrange_row_major_to_amp_f16_o1_r64_p64_c72_p80-:-z1211Z5z*z*breasonOperatorInputsvalue8kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(TransposedRight), tiling: TensorTiling { tile_count: 192, replicas: 1, axes: [AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(2), partitions: 12, block_size: 64, padding_multiple: 64, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements4608input0Elements4608invocations1 bJ7Moperation.5exchange :z2breasonOperatorInputsvalue9invocations1daB#eFl'>ibJ+   m  N  /  qR3 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)2)B)Z-b1reasonLocalCopyvalue10bytes10240patternContiguousinvocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  MM)7*744>!?!IISS]|^|hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bWuoperation.6ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue10bytes2048patternContiguousinvocations5 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&rr\ \""$F%F'')0*0,,./11346y6y88:c;c==?M@MBBD7E7GGI!J!LLNOQzQzSSU`V`XXZF[F]]_,`,b]b]ddfAgAiik%l%nnpNqNssu2v2xxz{}}llބކRRʼnŋ88  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1! bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSWSLWSSTS!S!S5%S5%Sk:S(k:S~S~SSS/S(/SSS㩄S㩄SGSGSS SS S%SL %SmCS mCSSS SS\S \SASAS;S;SSS;SD;SSSMlS MlSSSqS0qSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 aB bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 aB bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 aB  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`a)4XxA:^h%@dd Tx4Fj4 b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$mGk!EhBf@c bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1I[$ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  QQ)A*A44>1?1IIST^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaQQ A A  11!! y y""$i%i'')Y*Y,,.I/I113949668)9);;=>@@B C EEGGIqJqLLN]O]QQSITIVVX5Y5[[]!^!``b c eeggiojolln[o[qqsGtGvvx3y3{{}~ mmYYϑϓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSw˃Sw˃SSLS%ST%S!S!S;%S;%S:S(:SSSSSSS5S(5SSSgSgSSSS SS S+SL +SsCS sCSSS SS\S \SSSgSgSOSOS/SD/SSSAlS AlSYSYS)S0)SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 a5F bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 a5F bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1^  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 a5F  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$b$H:&&J(~(L2V**N Nr b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$&(*, . 02 bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1]:t } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  RR)@*@44>0?0IIST^^histst} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddRR @ @  ..  o o""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGIXJXLLNFOFQQS4T4VVX"Y"[[]^``bbdueuggicjcllnLoLqqs0t0vvxy{{}}jj܂܄NN22 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leSʃSʃSWSLWSSTS!S!S5%S5%Se:S(e:S~S~SSS/S(/SSSݩSݩSASASS SS S%SL %SLS LSsRS sRS[S [SSSSSS@S#S#SSS5lS5lS)St)S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 aB bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 aB bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b   b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 aB  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b)goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1D$$HB8(L(|2V<` b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b)goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1$=a; ^ 8 \  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1m6H } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  MM)9*944>%?%IIS T ^^hhrisi} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%ss_ _""$K%K'')7*7,,.#/#11346688:q;q==?]@]BBDIEIGGI5J5LLNOQQSTVyVyXXZa[a]]_I`Ibbd1e1ggijllnoququssu]v]xxzE{E}}--qqYY b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1leS}˃S}˃SSLS%ST%S!S!S;%S;%S:S(:SSSSSSS5S(5SSSgSgSSSS SS S+SL +SsCS sCS=RS =RSy[S y[SSSwSwSSSSDSSS/lS /lS S S߉S0߉SwS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 a5G bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 a5G bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 a5G  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$b$H &n&J(x(L!**N+ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@4 6  8  :  <  > @B bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1Q. hE } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  LL)6*644>"?"IIS T ^~^~hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbLL 6 6      iiS S""$=%='')'*',,./11335p6p88:Z;Z==?D@DBBD.E.GGIJLLNOQyQySSUcVcXXZM[M]]_7`7bbd!e!ggi j llnnpdqdssuHvHxxz,{,}}ff؉؋JJ.. b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS]SL]S%ST%S!S!S;%S;%Sq:S(q:S~S~SSS5S(5SSS驄S驄SMSMSS SS S+SL +SsCS sCS]SS ]SS\S \SᛅSᛅS鮅S鮅S1S1S5SD5S S SGlS GlSkSkS;S0;SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 cD bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 cD bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 cD  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`a&&J3  DDhyk"FFj=$HHlu#G b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@&zjZ~Jn:^*N> . bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1V*~R } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  OO)=*=44>+?+IIST^^hhrtst} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaOO = =  ++~~l l""$Z%Z'')H*H,,.6/6113$4$6689;;=>@w@wBBDeEeGGISJSLLN=O=QQS'T'VVXY[[]]_p`pbbdZeZggiDjDlln.o.qqstvvxy{w{w}}aaւքKK55 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSq˃Sq˃SSLS1ST1S!S!SG%SG%S:S(:SGSGSSSAS(ASSS[S[SSS!S !SS S7SL 7SCS CS?SS ?SS{\S {\SÛSÛSSSSS5SD5S S SGlS GlSqSqSAS0ASS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 ahQ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 ahQ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 ahQ  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`a&&J />>b_2VVz -Jnn]+ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$<0THl`x(@4X bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1hP8  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  \\)H*H44>4?4IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hh\\ P P  DD88,,   "#%%'(*v*v,,.b/b113N4N668:9:;;=&>&@@BCEEGGItJtLLN\O\QQSDTDVVX,Y,[[]^``bbdpepggiXjXlln@o@qqs(t(vvxy{{}}llTTȇȉ<<$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lkSʃSʃS]SL]S%ST%S!S!S;%S;%Sk:S(k:S~S~SSS5S(5SSS㩄S㩄SGSGSS SS S+SL +SLS LSRS RS[S [SSS!S!SWS@WS#S#SSS5lS5lSStS߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 bC bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 bC bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1c  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 bC  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b)goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1D$$H(L&J<`(LPt*Ne b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b)goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1$r b R v B f  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1&zN"v } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  NN)8*844>"?"IIST^{^{hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbNN : :  &&tt` `""$K%K'')5*5,,./113 4 6~6~88:h;h==?R@RBBD>b2VVz -Jnn= b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@L p d  |  , D 8\Pth bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1pX@( } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  VV)>*>44>&?&IIS T ^|^|hhr`s`} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&& ~ ~""$j%j'')R*R,,.:/:113"4"668 9 ;~;~==?f@fBBDNENGGI6J6LLNOQQSSUpVpXXZT[T]]_8`8bbdeggijlrlrnnpVqVssu:v:xxz{}}ttXXʉʋ<<  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃScSLcSCSTCS!S!SY%SY%Sq:S(q:S~S~SSSSS(SSєSєS驄S驄SMSMS3S 3SS SISL ISCS CSRS RS[S [S'S'S%S%SCSCS5SD5S S SGlS GlSMSMSS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  yZ bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  yZ bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  yZ  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H(L 00TTx<`DhhPtX||d b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$|tl d\TxLpDh bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1X0` } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  UU)A*A44>+?+IIST^^hhroso} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeYY M M  AA55%%  ""$s%s'')_*_,,.K/K113747668#9#;;=>@@BBDkEkGGISJSLLN;O;QQS#T#VVX Y [[]]_g`gbbdOeOggi7j7llnoqqstv{v{xxzc{c}}KK33ss b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSk˃Sk˃SSLS+ST+S!S!SA%SA%S:S(:SASASSS;S(;SSSUSUSSSS SS S1SL 1SyCS yCSRS RS[S [SSSSSSS#SD#SSS5lS 5lS S S߉S0߉SqS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|H b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 a5E bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|H b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 a5E bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|H  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 a5E  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H>.RRv$H \.R;8\!EEi b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$.R\!EOs }8Bf bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1sfYL } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  PP):*:44>"?"IIST^{^{hhrasa} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))uu_ _""$I%I'')3*3,,./11346|6|88:f;f==?N@NBBD4E4GGIJLLNOQsQsSSUYVYXXZ?[?]]_%`%bbd e g~g~iikdldnnpJqJssu0v0xxz{}}ooUUljNj99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃS]SL]S=ST=S!S!SS%SS%Sq:S(q:S~S~SSSMS(MS˔S˔S驄S驄SMSMS-S -SS SCSL CSLS LSgRS gRS[S [SSSSS{S@{SSSSS/lS/lS St S߉S߉SMS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1  xY bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1  xY bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1  xY  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b)goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1D$$H 08\:Lp D`*N b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b)goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1$<`4 X , P $ H  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations18h@ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  QQ)9*944>?IIST^u^uhhrYsY} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%%m m""$U%U'')=*=,,.%/%113 4 6688:i;i==?O@OBBD3E3GGIJLLNNPmQmSSUQVQXXZ5[5]]_`bbddfogoiikSlSnnp7q7ssuvxxzz|q}qUUDŽdž99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSk˃Sk˃SSLS+ST+S!S!SA%SA%S:S(:SGSGSSS;S(;SSS[S[SSSS SS S1SL 1SyCS yCSRS RS[S [SSS)S)SqSqSSDSSS#lS #lS S Sc߉S0c߉S S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 a5F bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 a5F bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1L  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 a5F  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H /.RRv; \!E;+O!EEi b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@p  + 5 Y c  (LVz?Im bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1?2%  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  NN)4*444>?IISS]k^khhrMsM} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))qqW W""$=%='')#*#,,. / 1|1|335b6b88:H;H==?,@,BBDEGGIIKaLaNNPCQCSSU%V%XXZ[]x]x__aZbZddf;?;IIS'T'^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeWW I I  ;;--  "#%|%|'')n*n,,.`/`113R4R668D9D;;=6>6@@B(C(EEGHJJL M OOQQSoToVVX]Y][[]K^K``b7c7eeg!h!jjl m o~o~qqsdtdvvxJyJ{{}0~0nnRRđē b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqS_˃S_˃SSLS+ST+S!S!SA%SA%S:S(:S;S;SSS;S(;SSSOSOSSSS SS S1SL 1SyCS yCSSS SS\S \S/S/SeSeSSS5SD5S S SGlS GlSASASS0SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q= b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 aaC bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q= b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 aaC bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q=  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 aaC  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H`Dhh 0  .d*NNr4nh b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$Dh *Nn4Tx:^ ~  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1~|zx } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  UU)?*?44>*?*IIST^^hhrpsp} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhXX H H  88((t t""$^%^'')H*H,,.2/211346689;{;{==?e@eBBDOEOGGI:J:LLN"O"QQS T V~V~XXZf[f]]_N`Nbbd6e6ggijllnoqzqzssubvbxxzJ{J}}22zzdd b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSʃSʃScSLcS+ST+S!S!SA%SA%Sw:S(w:S~S~SSS;S(;SSS梅S梅SSSSSS SS S1SL 1S LS  LSSS SS?\S ?\SSSSSS@S5S5S S SGlSGlS/St/S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 bC bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 bC bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 bC  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b)goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1D$$H:^Jnhp"'*N b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b)goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1$p ~  )  7 ! E  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1oR5{ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>/?/IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIWJWLLNAOAQQS+T+VVXY[[]]_t`tbbdZeZggi@j@lln$o$qqstvwvwxxzY{Y}};;qqSS b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lqSe˃Se˃SSLS+ST+S!S!SA%SA%S:S(:SASASSS;S(;SSSUSUSSSS SS S1SL 1SyCS yCSRS RS \S  \SQSQSSSSS)SD)SSS;lS ;lS S S߉S0߉SwS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uA b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 aaC bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uA b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 aaC bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1uA  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 aaC  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H+ODhhCg  .[*NNr sn b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ D d  * J n 0Tt:Z~  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1vtrpn  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  QQ);*;44>%?%IIS T ^|^|hhrbsb} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **vv` `""$J%J'')4*4,,./11346}6}88:g;g==?Q@QBBD;E;GGI%J%LLN O Q~Q~SSUdVdXXZJ[J]]_0`0bbdeggiikjljnnpNqNssu2v2xxz{}}llބކRRʼnŋ88 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwS˃S˃SiSLiS1ST1S!S!SG%SG%S}:S(}:S~S~SSSAS(ASSSSSYSYS!S !SS S7SL 7SCS CS?SS ?SS{\S {\SÛSÛSuSuSSSMSDMS%S%S_lS _lSSSS0S+S4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 bC bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 bC bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 bC  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$HFj0TTxx8 D(LLp \ b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$$HHll""FFjj  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1n\J8 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  UU)E*E44>5?5IIS!T!^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%% } }""$m%m'')]*],,.M/M113=4=668-9-;;=>@@B C EEGGIuJuLLNaOaQQSMTMVVX9Y9[[]%^%``bceeggisjslln_o_qqsKtKvvx7y7{{}#~#qq]]ӑӓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwSY˃SY˃SSLS1ST1S!S!SG%SG%S:S(:S5S5SSSAS(ASSSISISSS!S !SS S7SL 7SCS CS?SS ?SS{\S {\SÛSÛS S SSSASDASSSSlS SlSqSqSAS0ASS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xf2 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 a7B bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xf2 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 a7B bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1xf2  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 a7B  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$HImZ~~ ++O#Gb_33W! b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$Z~+a2h9o@v # G  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1 $ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  VV)D*D44>4?4IIS"T"^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhVV D D  22  s s""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGI\J\LLNJOJQQS8T8VVX&Y&[[]^``bceyeyggigjgllnPoPqqs4t4vvxy{{}}nnRRćĉ66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwS˃S˃SiSLiS1ST1S!S!SG%SG%S}:S(}:S~S~SSSAS(ASSSSSYSYS!S !SS S7SL 7SLS LSSS SSK\S K\S'S'SSSMS@MS;S;SSSMlSMlSAStASSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 bC bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 bC bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1f  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 bC  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b)goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1D$$H=a\m)0T5Y b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b)goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1$ D D h h   B  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1&p^ } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  QQ)=*=44>)?)IIST^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))wwc c""$O%O'');*;,,.'/'11346688:u;u==?a@aBBDMEMGGI9J9LLN!O!QQS T V}V}XXZe[e]]_M`Mbbd5e5ggijllnoqyqyssuavaxxzI{I}}11uu]] b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1lwS_˃S_˃SSLS1ST1S!S!SG%SG%S:S(:S;S;SSSAS(ASSSOSOSSS!S !SS S7SL 7SCS CS!SS !SS]\S ]\SSSۼSۼS=S=S/SD/SSSAlS AlSSS߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zj6 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 a6B bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zj6 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 a6B bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1zj6  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 a6B  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$HBfVz~%v+O4W{Dh&2V b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ } * N 1U8\ ?cFj bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1-6?HQ  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  PP):*:44>&?&IIST^^hhrfsf} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffPP : :  $$mmW W""$A%A'')+*+,,./11335t6t88:^;^==?H@HBBD2E2GGIJLLNOQ}Q}SSUgVgXXZQ[Q]]_;`;bbd%e%ggijllnnphqhssuLvLxxz0{0}}jj܉܋NN22 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l}S ˃S ˃SoSLoS7ST7S!S!SM%SM%S:S(:S~S~SSSGS(GSŔSŔSSS_S_S'S 'SS S=SL =SCS CSRS RS3\S 3\S{S{S#S#SSSGSDGSSSYlS YlS}S}SMS0MSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 bC bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 bC bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 bC  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H7[Bff_ //SUyy?B b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$:^t5Ko "F\3 W  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1yrkd } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  SS)A*A44>/?/IIST^^hisxsx} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeSS A A  //  p p""$^%^'')L*L,,.:/:113(4(6689;;=>@{@{BBDiEiGGIWJWLLNAOAQQS+T+VVXY[[]]_t`tbbd^e^ggiHjHlln2o2qqstvvxy{{{{}}eeڂڄOOćĉ99## b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l}SS˃SS˃SSLS7ST7S!S!SM%SM%S:S(:S/S/SSSGS(GSŔSŔSCSCSSS'S 'SS S=SL =SCS CSoSS oSS\S \SSSSS?S?SMSDMS%S%S_lS _lSSSSS0SSS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m[' b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 aaC bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m[' b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 aaC bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m['  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 aaC  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H<`qd)MMq) DQuu --Q b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@$p(LPt,x0T X |  4  bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1(<P } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  ``)L*L44>8?8IIS T ^^his|s|} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ll`` T T  HH<<00$$  "#%%'(*z*z,,.f/f113R4R668>9>;;=*>*@@BCEEGHJxJxLLN`O`QQSHTHVVX0Y0[[]^``bcetetggi\j\llnDoDqqs,t,vvxy{{}}ppXẊ̉@@(( b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l}S ˃S ˃SoSLoS7ST7S!S!SM%SM%S:S(:S~S~SSSGS(GSŔSŔSSS_S_S'S 'SS S=SL =SLS LS SS  SSE\S E\S!S!SSSS@S5S5S S SGlSGlSStS߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 bC bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 bC bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 bC  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b)goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1D$$H:^nf7 ](L& b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b)goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r9_p16_c72_p80-:-z1211Z5bYz]"]za"abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements648input0Elements648invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1$ m . D h  ~ ? bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1]VOHA  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  RR)<*<44>&?&IIS T ^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **xxd d""$O%O'')9*9,,.#/#113 4 6688:l;l==?V@VBBD@E@GGI*J*LLNOQQSSUiViXXZO[O]]_5`5bbdeggijltltnnpZqZssu@v@xxz&{&}} ee؉؋KK11 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations1l}SY˃SY˃SSLS7ST7S!S!SM%SM%S:S(:S/S/SSSGS(GSŔSŔSCSCSSS'S 'SS S=SL =SCS CScSS cSS\S \S盅S盅S7S7SSS5SD5S S SGlS GlS S S߉S0߉SS4} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o_+ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.0exchange :z2breasonOperatorInputsvalue4invocations1 aaC bwoperation.0ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue4kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.1ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue5bytes2560patternContiguousinvocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o_+ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7M7operation.1exchange :z2breasonOperatorInputsvalue5invocations1 aaC bwoperation.1ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue5kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1  bWuoperation.2ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue6bytes2560patternContiguousinvocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1o_+  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1  bJ7M7operation.2exchange :z2breasonOperatorInputsvalue6invocations1 aaC  bwoperation.2ipu_stack_reduce_sum_f165:5z9299Z=2MzQ*QzU*UzY*Yb]reasonOperatorKernelvalue6kernelSpecPlanned(ReductionSum { partials: 2 })outputElements1280input0Elements1280input1Elements1280invocations1 bWuoperation.3ipu_stack_copy_u64%:%R)2))2)*)B)Z-b1reasonLocalCopyvalue7bytes2560patternContiguousinvocations1 b-goperation.3ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue7kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1`$$H?clo"FFj+ 7[DhhB b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1 b-goperation.4ipu_stack_unpack_amp_to_row_major_f16_o1_r16_p16_c72_p80-:-z1211Z5bYz]*]za*abereasonOperatorKernelvalue8kernelSpecPlanned(Rearrange { from: Layout { order: Amp(TransposedLeft), tiling: TensorTiling { tile_count: 736, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 46, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 16, padding: Zero, tile_stride: Some(46) }] }, memory_class: Ipu21Interleaved }, to: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard } })outputElements1152input0Elements1152invocations1- bJ7Moperation.4exchange :z2breasonOperatorInputsvalue8invocations1@ 8 \  `<@dh DHl bJ7MWoperation.5exchange :z2breasonOperatorInputsvalue9invocations1dx  4  } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  ZZ)B*B44>*?*IIST^^hhrdsd} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jjZZ J J  ::**   ""$n%n'')V*V,,.>/>113&4&6689;;==?j@jBBDRERGGI:J:LLNOQQSTVtVtXXZX[X]]_<`v>xxz"{"}}xx\\Ή΋@@$$ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LS)˃S)˃SSSS$S:S:S~S~SmS$mSSSwSwSS$SRSRS[S4[SɧST ɧSq S q SYS YS1S 1SklS klSqS qSASASS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations14y& b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations14y& b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations14y& b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 a;;__  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10zz  jj)V*V44>>?>IIS&T&^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1zznn b b  VVJJ::&&  "#%%'')t*t,,.`/`113L4L668898;;=">"@@B C E~E~GGIfJfLLNNONQQS6T6VVXY[[]^`z`zbbdbebggiJjJlln2o2qqstvvxy{v{v}}^^҂҄FF++ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LSM˃SM˃SS SSS:S:S)S)S=S=S=Sp=SSSSSRS`RS!\S!\SɧS$ ɧSq S q SAS ASS SSlS SlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1aO b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1M} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1aO b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1M} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1aO b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1M k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 a/Sc!  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10rr  ]])G*G44>-?-IIST^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1rr^^ J J  66""  l l""$V%V'')@*@,,.*/*11346688:s;s==?Y@YBBD?E?GGI%J%LLN O Q~Q~SSUdVdXXZJ[J]]_0`0bbdeggiikolonnpUqUssu;v;xxz!{!}}zz^^ЉЋBB&& b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LS#˃S#˃SSSS$S:S:S~S~SmS$mSSSwSwSS$S9SS9SSu\S4u\SɧST ɧSq S q SAS ASS SSlS SlS S  S߉S߉SqS} bJ7UGoperation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations14w" b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7UGoperation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations14w" b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7UGoperation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations14w" b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 a+SwEi  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10xx  dd)L*L44>0?0IIST^^hhrhsh} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1xxhh X X  HH88$$   ""$h%h'')P*P,,.8/8113 4 6689;|;|==?`@`BBDDEDGGI(J(LLN O Q~Q~SSUbVbXXZF[F]]_*`*bbdeggiikdldnnpHqHssu,v,xxz{}}ff؄؆HH**  b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LSS˃SS˃SS SSS:S:S)S)S7S7S=Sp=SSSSSRS`RS3\S3\SɧS$ ɧSq S q S/S /SS SAlS AlS S  S߉S߉S)S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1aQ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1L} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1aQ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1L} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1aQ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1L k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 a!5Ym -  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  ]])C*C44>%?%IIST^x^xhhrXsX} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1tt`` L L  88$$  f f""$L%L'')2*2,,./11335q6q88:W;W==?9@9BBDEGGIIKnLnNNPPQPSSU2V2XXZ[]]__agbgddfIgIiik+l+nnp q s~s~uuw`x`zz|B}B$$ttTTĐĒ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S˃S˃SS SSS:S:S~S~SgSgS Sp SqSqSSSgRS`gRS[S[SɧS$ ɧSq S q S_S _S7S 7SqlS qlSS SSS1S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1v# b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1v# b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1v# b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 br .>  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10vv  hh)Z*Z44>J?JIIS6T6^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1vvhh Z Z  LL>>00""  "#%%'(**,,.q/q113c4c668U9U;;=G>G@@B9C9EEG+H+JJLMOOQRT~T~VVXlYl[[]Z^Z``bFcFeeg0h0jjlmooqrtstsvvxYyY{{}?~?%% {{__ёѓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}SS˃SS˃SS SSS:S:S/S/SISISCSpCSSSSSRS`RS[S[SɧS$ ɧSq S q SGS GSS SYlS YlSSS SS#S#SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1jX$ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1a} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1jX$ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1a} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1jX$ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1a k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 a51UQu  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  ^^)H*H44>1?1IIST^^hisusu} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qqaa Q Q  AA11 } }""$g%g'')Q*Q,,.;/;113%4%6689;;==?n@nBBDXEXGGIAJALLN)O)QQSTVVXXZm[m]]_U`Ubbd=e=ggi%j%lln o qqssuivixxzQ{Q}}99!! ii b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S˃S˃SS SSS:S:S~S~SgSgSSpSkSkSSS'SS`'SSc\Sc\SɧS$ ɧSq S q SGS GSS SYlS YlSGS GSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1t b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 a)Mq9]  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10tt  bb)P*P44>>,,  ""$m%m'')[*[,,.I/I113747668%9%;;=>@@BCExExGGIdJdLLNNONQQS8T8VVX"Y"[[] ^ ``bbdgegggiMjMlln1o1qqstvvxxzf{f}}HH** ||^^ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}SY˃SY˃SS SSS:S:S/S/SISISCSpCSSSSSQSS`QSS\S\SɧS$ ɧSq S q S;S ;SS SMlS MlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1jZ& b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1a} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1jZ& b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1a} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1jZ& b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1a k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 a /_ -]+  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10qq  \\)F*F44>.?.IIST^^hhrisi} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1qq]] I I  55!!  k k""$U%U'')?*?,,.)/)11346688:r;r==?\@\BBDFEFGGI.J.LLNOQQSSUmVmXXZS[S]]_9`9bbdeggijlslsnnpWqWssu;v;xxz{}}uuYỶ̋??%% b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S˃S˃S{S {SSS:S:S~S~SmSmSSpSkSkSSSORS`ORS[S[SçS$ çSk S k S_S _S7S 7SqlS qlSS SSS=S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1s  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1s  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1s  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 a ku(  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10rr  bb)R*R44>@?@IIS,T,^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1rrbb R R  BB22""  "#%z%z'')j*j,,.Z/Z113J4J668:9:;;=*>*@@BCEEG H JJLLNlOlQQSXTXVVXDYD[[]0^0``bceeghj~j~llnjojqqsVtVvvxByB{{}.~.||hhޑޓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}SY˃SY˃SS SSS:S:S5S5SSSISpISSSSSORS`ORS[S[SçS$ çSk S k SMS MS%S %S_lS _lSwS wSGSGSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1sa- b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations15} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1sa- b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations15} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1sa- b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations15 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 a ku(  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  [[)I*I44>7?7IIS%T%^^h i s{s{} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mm[[ I I  77%% x x""$f%f'')T*T,,.B/B1130406689;;= > @@BBDqEqGGI_J_LLNMOMQQS;T;VVX)Y)[[]^``bce|e|ggijjjllnSoSqqs7t7vvxy{{}}qqUULJlj99 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}S˃S˃S{S {SSS:S:S~S~SsSsSSpSeSeSSS3SS`3SSo\So\SçS$ çSk S k SGS GSS SYlS YlSMS MSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1q b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 a+SwEi  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10pp  \\)H*H44>2?2IIST^^hisvsv} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1pp\\ H H  44    n n""$Z%Z'')F*F,,.2/21134668 9 ;;==?l@lBBDXEXGGIBJBLLN*O*QQSTVVXXZn[n]]_V`Vbbd>e>ggi&j&llnoqqssujvjxxzR{R}}::"" ~~ff b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17L}SY˃SY˃SS SSS:S:S5S5SSSISpISSSSS]SS`]SS\S\SçS$ çSk S k S;S ;SS SMlS MlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1sc/ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations15} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1sc/ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations15} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1sc/ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations15 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 a 1e5i9  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  WW)A*A44>+?+IIST^^hhrksk} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mmWW A A  ++tt^ ^""$H%H'')2*2,,./11346{6{88:e;e==?O@OBBD9E9GGI#J#LLN O QQSSUlVlXXZV[V]]_@`@bbd*e*ggijllnnpmqmssuQvQxxz5{5}}ooSSŎŐ77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS˃S˃SuS uSSS:S:S~S~SaSaSSpSeSeSSSIRS`IRS[S[SS$ Se S e SMS MS%S %S_lS _lSS SSSSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1p b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 a lv)  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  \\)J*J44>6?6IIS T ^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nn\\ J J  88&& y y""$g%g'')U*U,,.C/C1131416689;;= > @@BBDrErGGI^J^LLNHOHQQS2T2VVXY[[]^`{`{bbdeeeggiOjOlln9o9qqs#t#vvx y {{}}llVVˇˉ@@** b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS_˃S_˃SS SSS:S:S;S;SISISOSpOSSSSSIRS`IRS[S[SS$ Se S e SGS GSS SYlS YlS}S }SMSMSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|j6 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1j} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|j6 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1j} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|j6 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1j k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 a kw *  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  aa)M*M44>7?7IIST^^his{s{} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mmaa U U  II==11%%  "#%%'(*{*{,,.g/g113S4S668?9?;;=+>+@@BCEEGHJwJwLLN_O_QQSGTGVVX/Y/[[]^``bbdsesggi[j[llnCoCqqs+t+vvxy{{}}ooWWˇˉ??'' b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS ˃S ˃SoS oSSS:S:S~S~SaSaSSpS_S_SSS!SS`!SS]\S]\SS$ Se S e S5S 5S S  SGlS GlSS S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1n b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 a)Mq:^  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  YY)C*C44>+?+IIST^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mmYY E E  11  k k""$V%V'')@*@,,.*/*11346688:s;s==?]@]BBDGEGGGI/J/LLNOQQSSUnVnXXZT[T]]_:`:bbd e ggijlylynnp_q_ssuEvExxz+{+}}jj݉݋PPÎÐ66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LwS_˃S_˃SS SSS:S:S;S;SOSOSOSpOSSSSSKSS`KSS\S\SS$ Se S e S/S /SS SAlS AlSS S߉S߉S}S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|l8 b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1k} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|l8 b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1k} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|l8 b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1k k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 a /_ -]+  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10mm  ]])E*E44>+?+IIST^^hhrese} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1mm]] M M  ==--   ""$q%q'')Y*Y,,.A/A113)4)6689;;==?m@mBBDUEUGGI;J;LLNOQQSTVuVuXXZY[Y]]_=`=bbd!e!ggijlwlwnnp[q[ssu?v?xxz#{#}}yy]]ωϋAA%% b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS ˃S ˃SoS oSSS:S:S~S~SmSmSSpS_S_SSSmRS`mRS[S[SS$ Se S e S;S ;SS SMlS MlSYS YS)S)SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1m b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  0: D*N  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10nn  ^^)J*J44>2?2IIST^^histst} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1nnbb V V  JJ>>..  "#%|%|'')h*h,,.T/T113@4@668,9,;;=>@@BBDrErGGIZJZLLNBOBQQS*T*VVXY[[]]_n`nbbdVeVggi>j>lln&o&qqstvvxxzj{j}}RRƂƄ::xx b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSe˃Se˃SS SSS:S:S;S;SSSOSpOSSSSSmRS`mRS[S[SS$ Se S e S#S #SS S5lS 5lS S  S߉S߉SqS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1s? b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations16} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1s? b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations16} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1s? b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations16 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  0: D*N  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  QQ);*;44>!?!IIST^z^zhhr^s^} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffRR > >  **vv` `""$J%J'')4*4,,./11346}6}88:g;g==?M@MBBD3E3GGIJLLNNPrQrSSUXVXXXZ>[>]]_$`$bbd e g}g}iikclcnnpIqIssu/v/xxz{}}nnRRĉċ66 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS˃S˃SiS iSSS}:S}:S~S~SmSmSSpSYSYSSSyRS`yRS[S[SS$ Se S e S#S #SS S5lS 5lS S  S߉S߉SYS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1k b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1k b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1k b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 A'K1U;_  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ll  XX)@*@44>$?$IIST^z^zhhr\s\} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ll\\ L L  <<,, t t""$\%\'')D*D,,.,/,11346688:p;p==?T@TBBD8E8GGIJLLNOQrQrSSUVVVXXZ:[:]]_`bbdegtgtiikXlXnnp?IISS]l^lhhrLsL} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhTT @ @  ,,ttZ Z""$@%@'')&*&,,. / 11335e6e88:K;K==?-@-BBDEGGIIKbLbNNPDQDSSU&V&XXZ[]y]y__a[b[ddf=g=iiklnnpqsrsruuwTxTzz|6}6hh؋؍HH b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqS˃S˃SiS iSSSw:Sw:S~S~S[S[S梅Sp梅SSSSSSSmRS`mRS[S[SS$ S_ S _ SGS GSS SYlS YlSS SqSqSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1j b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 5?%I/S  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10jj  \\)N*N44>>?>IIS*T*^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1jj\\ N N  @@22$$  "#%%'')s*s,,.e/e113W4W668I9I;;=;>;@@B-C-EEGHJJL M OOQQSrTrVVX`Y`[[]N^N``b:c:eeg$h$jjlmooqqsgtgvvxMyM{{}3~3ooSSőœ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSk˃Sk˃SS SSS:S:SASAS=S=SUSpUSSSSSgRS`gRS[S[SS$ S_ S _ S/S /SS SAlS AlS5S 5SSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|H b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1f} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|H b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1f} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1|H b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1f k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1  .8B(L  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  RR)<*<44>%?%IIS T ^^hhrisi} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeUU E E  55%%q q""$[%['')E*E,,.///11346689;x;x==?b@bBBDLELGGI5J5LLNOQQSTVyVyXXZa[a]]_I`Ibbd1e1ggijllnoququssu]v]xxzE{E}}--ss]] b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSʃSʃScS cSSSw:Sw:S~S~S[S[S梅Sp梅SSSSSSS SS` SSE\SE\SS$ S_ S _ S/S /SS SAlS AlS/S /S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1h b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 =am)5Y  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10hh  VV)D*D44>0?0IIST^^hhrjsj} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1hhVV D D  22  s s""$a%a'')O*O,,.=/=113+4+6689;;=>@~@~BBDlElGGIXJXLLNBOBQQS,T,VVXY[[]^`u`ubbd[e[ggiAjAlln%o%qqstvxvxxxzZ{Z}}<<ppRR b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LqSk˃Sk˃SS SSS:S:SGSGSCSCS[Sp[SSSSSRS`RS[S[SS$ S_ S _ S#S #SS S5lS 5lS S  S߉S߉SkS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~J b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1g} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~J b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1g} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1~J b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1g k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 $H2V>bJn  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ee  PP):*:44>"?"IIST^y^yhhr]s]} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1eeQQ = =  ))uu_ _""$I%I'')3*3,,./11346|6|88:f;f==?P@PBBD:E:GGI"J"LLNOQ{Q{SSUaVaXXZG[G]]_-`-bbdeggiikglgnnpKqKssu/v/xxz{}}iiۄۆMM33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃScS cSSSq:Sq:S~S~SaSaS驄Sp驄SMSMSSSmRS`mRS[S[SS$ SY S Y SAS ASS SSlS SlSS S}S}S%S} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1g b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 <"F,P6Z  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10ff  VV)F*F44>4?4IIS T ^^hiss} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ffVV F F  66&& ~ ~""$n%n'')^*^,,.N/N113>4>668.9.;;=>@@BCEEGGItJtLLN`O`QQSLTLVVX8Y8[[]$^$``bceeggirjrlln`o`qqsLtLvvx8y8{{}$~$rr^^ԑԓ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSk˃Sk˃SS SSS:S:SGSGS S S[Sp[SSSSSgRS`gRS[S[SS$ SY S Y S;S ;SS SMlS MlSeS eS5S5SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations15} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations15} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Q b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations15 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 5?%I/S  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  OO)=*=44>+?+IIST^^hisysy} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaOO = =  ++~~l l""$Z%Z'')H*H,,.6/6113$4$6689;;=>@w@wBBDeEeGGISJSLLNAOAQQS/T/VVXY[[] ^ ``bbdpepggi^j^llnQoQqqs5t5vvxy{{}}ooSSŇʼn77 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSʃSʃS]S ]SSSq:Sq:S~S~SaSaS驄Sp驄SMSMSSSSS`SS\S\SS$ SY S Y S/S /SS SAlS AlS5S 5SSSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1 } bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1 } bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1e b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1  k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 ^ 0h;s  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10dd  PP)<*<44>&?&IIST^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1ddPP < <  ((vvb b""$N%N''):*:,,.&/&11346688:t;t==?`@`BBDLELGGI6J6LLNOQQSTVzVzXXZb[b]]_J`Jbbd2e2ggijllnoqzqzssubvbxxzJ{J}}22vv^^ b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LkSq˃Sq˃SS SSS:S:SGSGS S S[Sp[SSSSS'SS`'SSc\Sc\SS$ SY S Y S)S )SS S;lS ;lS S  S߉S߉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations15} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations15} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1S b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations15 k bJ7Moperation.3exchange :z2breasonOperatorInputsvalue7invocations1 Ei}=Qu  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r8_p8_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements576input0Elements576invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  KK)5*544>?IIS T ^^hhrgsg} b)woperation.6ipu_stack_gemm_f16_init_large_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements6144input0Elements576input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_large_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements6272input0Elements6144invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaKK 5 5    ~~hhR R""$<%<'')&*&,,./11335o6o88:Y;Y==?C@CBBD-E-GGIJLLNOQvQvSSU`V`XXZJ[J]]_4`4bbdeggijl}l}nnpiqissuMvMxxz1{1}}kk݉݋OO33 b)woperation.6ipu_stack_gemm_f16_init_large_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements640input0Elements6144input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_large_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements576input0Elements640input1Elements6272invocations17LeSʃSʃSWS WSSSk:Sk:S~S~SOSOS㩄Sp㩄SGSGSSSRS`RSO[SO[SS SY S| Y S;S ;Sc~S c~SANS ANSqS qSgމSgމSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1d b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations17A?c  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10bb  PP)@*@44>,?,IIST^^h i ss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1bbPP > >  ,,m m""$]%]'')K*K,,.9/9113'4'6689;;=>@z@zBBDhEhGGITJTLLN>O>QQS(T(VVXY[[]]_q`qbbd[e[ggiEjElln;o;qqs%t%vvxy{{}}nnXX͇͉BB,, b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LeSq˃Sq˃SS SSS:S:SMSMS7S7SaSpaSSSSSRS`RS=[S=[SS SY S| Y SAS ASi~S i~SGNS GNSwS wSmމSmމSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1d} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1d} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1Z b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1d k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1: D9]  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  UU)K*K44>5?5IIST^^hiss} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaUU I I  ==11%%  " # %%'(*y*y,,.e/e113Q4Q668=9=;;=)>)@@BCEEGHJuJuLLN]O]QQSETEVVX-Y-[[]^``bbdqeqggiYjYllnMoMqqs5t5vvxy{{}~yyaaՇՉII11 b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LeSʃSʃSWS WSSSe:Se:S~S~SOSOSݩSpݩSASASSSSS`SS\S\SS SY S| Y S)S )SQ~S Q~S/NS /NS S  SމSމSS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b  b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b  b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1b  b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}"}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements648input0Elements41472input1Elements5184invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1 k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1Mq9m  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  MM);*;44>#?#IIS T ^^hhrnsn} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaMM 9 9  %%ss_ _""$N%N'')8*8,,."/"113 4 6688:k;k==?U@UBBD?E?GGI'J'LLN O QQSSUfVfXXZL[L]]_2`2bbdeggiikqlqnnpcqcssuIvIxxz/{/}}nnTTǎǐ:: b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations17LeSw˃Sw˃SS SSS:S:SMSMS7S7SaSpaSSSSS!SS`!SS]\S]\SS SY S| Y S)S )SQ~S Q~S/NS /NS S  S݉S݉SS} bJ7U'operation.0exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.0ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue4kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.0exchange :z2breasonOperatorInputsvalue4invocations1e} bJ7U'operation.1exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.1ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue5kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.1exchange :z2breasonOperatorInputsvalue5invocations1e} bJ7U'operation.2exchange :!2!!b%reasonOperatorInput { input: 0 }value0invocations1\ b-woperation.2ipu_stack_gemm_f16_init_small_rows_interleaved_k576_c16_r80_r805:5z9299Z=yz}*}z2z*breasonOperatorKernelvalue6kernelSpecPlanned(Gemm { multiply: F16, accumulate: F16, mode: Initialize, weights: Interleaved, inner_block: 576, output_columns: 16 })outputElements1152input0Elements41472input1Elements9216invocations1 bJ7Moperation.2exchange :z2breasonOperatorInputsvalue6invocations1e k bJ7Mwoperation.3exchange :z2breasonOperatorInputsvalue7invocations1l"F<`  b-g operation.3ipu_stack_rearrange_row_major_to_amp_f16_o0_r7_p7_c72_p80-:-z1211Z5z"z"breasonOperatorInputsvalue7kernelSpecPlanned(Rearrange { from: Layout { order: RowMajor, tiling: TensorTiling { tile_count: 1, replicas: 1, axes: [] }, memory_class: Ipu21Standard }, to: Layout { order: Amp(Left), tiling: TensorTiling { tile_count: 1472, replicas: 1, axes: [AxisTiling { axis: FromEnd(2), partitions: 92, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(16) }, AxisTiling { axis: FromEnd(3), partitions: 16, block_size: 1, padding_multiple: 1, shard_padding_multiple: 1, padding: Reject, tile_stride: Some(1) }, AxisTiling { axis: FromEnd(1), partitions: 1, block_size: 16, padding_multiple: 16, shard_padding_multiple: 1, padding: Zero, tile_stride: None }] }, memory_class: Ipu21Standard } })outputElements504input0Elements504invocations1  b*7Ioperation.4sync :2breasonExchangeBarriervalue8invocations1  b*7Ioperation.5sync :2breasonExchangeBarriervalue9invocations1 } bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations10aa  QQ)A*A44>'?'IIS T ^^hhrmsm} b)woperation.6ipu_stack_gemm_f16_init_small_rows_k80_c768_r7_r85:5z9299Z=yz}*}z"z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 80, output_columns: 768 })outputElements5376input0Elements504input1Elements52488invocations1 b)goperation.6ipu_stack_attention_softmax_small_query_small_key_f16-:-z1211Z5aze*ezi*ibmreasonOperatorKernelvalue10kernelSpecPlanned(AttentionSoftmax { head_dimension: 72, key_columns: 729, padded_key_columns: 768 })outputElements5488input0Elements5376invocations1 bJ7Moperation.6exchange :z2breasonOperatorInputsvalue8invocations1aaQQ A A  11!! y y""$m%m'')U*U,,.=/=113%4%668 9 ;;==?i@iBBDQEQGGI7J7LLNOQQSSUqVqXXZU[U]]_9`9bbdeggijlslsnnpcqcssuGvGxxz+{+}}ee׉׋II-- b)woperation.6ipu_stack_gemm_f16_init_small_rows_k768_c80_r7_r85:5z9299Z=yz}"}z*z2breasonOperatorKernelvalue10kernelSpecPlanned(Gemm { multiply: F16, accumulate: F32, mode: Initialize, weights: Standard, inner_block: 768, output_columns: 80 })outputElements560input0Elements5376input1Elements52488invocations1 bR%woperation.6ipu_stack_attention_merge_small_query_f165:5z9299Z=2}z"z"z*breasonOperatorKernelvalue10kernelSpecPlanned(AttentionMerge { value_dimension: 72, padded_value_dimension: 80, key_block_columns: 768, initial: true, final_block: true })outputElements504input0Elements560input1Elements5488invocations1